170.210.53.25170.210.53.25/ms_idei/publicaciones/wp-content/uploads/2016/06/t… · - i -...
TRANSCRIPT
- i -
Innovative strategies in unstable environments: the case of
Argentinean firms
Ph.D. Thesis
Aalborg Universitet – Institut for Økonomi og Ledelse (Danmark)
Universidad Nacional de Quilmes – Departamento de Ciencias Sociales (Argentina)
Aalborg Universitet – Institut for Økonomi og Ledelse (Danmark)
Universidad Nacional de Quilmes – Departamento de Ciencias Sociales (Argentina)
PhD candidate: Diana Suárez
Supervisors AAU: Björn Johnson & Bengt-Åke Lundvall Supervisors UNQ: Bernardo Kosacoff & Gabriel Yoguel
July 2013
- i -
Innovative strategies in unstable environments:
the case of Argentinean firms
PhD Thesis
PhD candidate: Diana Suárez Supervisors AAU: Björn Johnson & Bengt-Åke Lundvall Supervisors UNQ: Bernardo Kosacoff & Gabriel Yoguel
Aalborg University—Business and Management Department (Denmark) National University of Quilmes—Social Sciences Department (Argentina)
July 2013
Summary
This thesis is about innovation, instability, and micro-heterogeneity. The theoretical
motivation behind the research is the lack of satisfactory explanations about the existence
of firms in Argentina that managed to cope much better than others firms with the
country’s instability and to compete based on innovation. The research question that has
guided this thesis is why some firms are better prepared to cope with unstable
environments than others.
The search for answers began with the selection of the National Innovation System (NIS)
approach as a theoretical framework. This choice was motivated by the fact that the
innovation system is defined by the interaction of organizations and institutions with
different characteristics. The interaction between the NIS as a whole and diverse firms is
seen as shaping the intensity and direction of innovation and technological change.
This study’s review of the NIS approach and its application to the Argentinean case
preliminary concludes that there is a missing link in its theoretical foundations to the extent
that the interactive relationship between the NIS and firm behavior is not properly
accounted for. In this sense, the concept of variety is not enough to explain the micro-
heterogeneity of the system, and the static characteristics the approach has acquired works
against the possibility of explaining how firms deal with change. In the search for answers
to the research question, this thesis will zoom in on the NIS approach in order to look at
Summary
- ii -
the firm level from a dynamic perspective and will propose the concept of strategy to
explain the dynamic relationship between the firm and the environment. This concept will
be empirically tested on a set of 800 manufacturing Argentinean firms from the period
1998–2006.
This analysis will lead to the preliminary conclusion that micro-heterogeneity is
endogenous to the dynamics of the NIS, within which firms will adjust their strategies
differently given changes in the environment and its capabilities. The results show that,
facing the same instable environment, Argentinean firms reacted differently in terms of
their innovative strategies. This provides evidence regarding the importance of studying
firm behavior from a dynamic perspective, allowing not only for the firm to modify the
way it seeks innovations, but also for the possibility of not seeking innovations at all. The
findings of this research suggest that the firm’s learning capabilities and innovation
investments are key elements to understanding why some firms are better prepared to deal
with unstable environments, which is not because they are more likely to successfully
innovate but because they have dealt better with the process of path creation. Firms with
an innovation-based strategy have accumulated the competences necessary for taking
advantage of environmental opportunities and avoiding environmental constraints. That is
why some firms are better prepared than others to deal with unstable environments.
Resumé
Denne afhandling handler om innovation, ustabilitet og heterogenitet på mikro-niveau.
Den teoretiske motivation for forskningen er manglen på en tilfredsstillende forklaring for,
hvordan nogle argentinske virksomheder klarede sig bedre gennem landets ustabilitet og
konkurrerer ved hjælp af innovation. Forskningsspørgsmålet, som har ledt denne
forskning, er, hvorfor nogle virksomheder er bedre til at klare sig i ustabile miljøer end
andre.
Søgen efter svar begyndte med valget af nationale innovationssystemer (NIS) som teoretisk
ramme. Dette valg var motiveret af det faktum, at et innovationssystem er defineret ved
interaktionen mellem organisationer og institutioner med forskellige karakteristika.
Interaktionen mellem NIS som helhed og forskelligartede virksomheder ses som givende
form og retning til innovation og teknologisk forandring.
Summary
- iii -
Studiets gennemgang af NIS-tilgangen og dennes anvendelse i det argentinske tilfælde
konkluderer indledningsvis, at der er et manglende led i tilgangens teoretiske fundament i
og med at det interaktive forhold mellem NIS og virksomheders ageren ikke er behørigt
inkluderet. I denne forstand er konceptets mangfoldighed ikke tilstrækkeligt til at forklare
systemets heterogenitet på mikro-niveau, og den statiske tendens, som tilgangen har
erhvervet sig, modvirker muligheden for at forklare, hvordan virksomheder reagerer på
forandring. I sin søgen efter svar til forskningsspørgsmålet zoomer afhandlingen ind på
NIS-tilgangen for at studere virksomhedsniveauet fra et dynamisk perspektiv. Afhandlingen
foreslår konceptet strategi til at forklare det dynamiske forhold mellem virksomheder og
deres omgivelser. Konceptet testes empirisk i en database af 800 argentinske
fremstillingsvirksomheder over perioden 1998-2006.
Analysen leder indledningsvis til konklusionen at heterogenitet på mikro-niveau er
endogent til dynamikken at et NIS, da virksomheder indenfor NISet tilpasser deres strategi
på forskellig vis for givne ændringer i miljø og dettes kapabiliteter. Resultater viser at, for et
givent ustabilt miljø, reagerede argentinske virksomheder forskelligt med hensyn til deres
innovationsstrategier. Dermed underbygges vigtigheden af, at studere virksomheders
strategier fra et dynamisk perspektiv, som tillader, at virksomheder ændrer måden hvorpå
de søger efter innovation, men også at virksomheder vælger ikke at søge efter innovation i
det hele taget. Resultatet af denne forskning antyder, at virksomheders læringsevne og
investeringer i innovation er kerneelementer i forståelsen af, hvorfor nogle virksomheder
bedre formår at administrere ustabile miljøer. Det skyldes ikke, at disse virksomheder har
større sandsynlighed for at lave succesfulde innovationer. Det skyldes snarer, at de er bedre
til at klare den proces der skaber udviklingsbaner. Virksomheder med en strategi baseret på
innovation har opbygget de kompetencer, som er nødvendige for at gribe muligheder i
miljøet og undgå begrænsninger i miljøet. Derfor er nogle virksomheder bedre til at klare
sig i ustabile miljøer end andre.
Resumen
La presente tesis tiene por objeto el análisis de la innovación, la inestabilidad y la micro-
heterogeneidad. La motivación teórica de esta investigación surge de la debilidad de las
explicaciones existentes respecto de la existencia, en Argentina, de firmas que han logrado
manejar mejor la inestabilidad del país y competir a partir de la innovación. La pregunta de
Summary
- iv -
investigación que guía esta tesis es: ¿por qué algunas firmas están mejor preparadas para
lidiar con un entorno inestable?
La búsqueda de respuestas comienza con la selección del enfoque de los Sistemas
Nacionales de Innovación (SNI) como marco teórico. Esta selección responde a que los
sistemas de innovación son definidos a partir de la interacción de diferentes organizaciones
e instituciones. De esta forma, la interacción al interior del SNI, incluidas las firmas,
determina la intensidad y dirección de la innovación y el cambio tecnológico.
La revisión del enfoque de los SNI hecha en esta investigación y su aplicación al caso
Argentino permite concluir de manera preliminar que existe un espacio vacío en sus
fundamentos teóricos en relación a la relación interactiva entre el sistema y las firmas. En
este sentido, el concepto de variedad no es suficiente para explicar la micro-heterogeneidad
del sistema, y las características estáticas que el enfoque ha adquirido atentan contra la
posibilidad de explicar cómo las firmas manejan el cambio. En la búsqueda de respuestas a
la pregunta planteada, esta tesis analiza la firma, dentro del enfoque de los SNI, desde una
perspectiva dinámica, y propone el concepto de estrategia para explicar la relación dinámica
entre la empresa y el entorno. El concepto es luego aplicado un conjunto de 800 firmas
manufactureras Argentinas para el período 1998-2006.
Las conclusiones preliminares del análisis empírico permiten sostener que la micro-
heterogeneidad es endógena al SNI, en el cual las firmas ajustarán sus estrategias de forma
diferente dados cambios en el entorno y dadas sus propias capacidades. Los resultados
muestran que enfrentadas al mismo entorno inestable, algunas firmas Argentinas
reaccionaron diferente en términos de su estrategia innovativa. Esto provee evidencia
respecto de la importancia de estudiar el comportamiento de las empresas desde una
perspectiva dinámica que permita no sólo que la firma modifique la forma en que busca
alcanzar innovaciones sino también la posibilidad de no buscar innovar. Los hallazgos de
esta investigación sugieren que las capacidades de aprendizaje y las inversiones en
innovación son elementos clave para comprender por qué algunas firmas están mejor
preparadas para lidiar con entornos inestables, lo que resulta no de las mayores
probabilidades de alcanzar innovaciones sino de la forma en que las firmas manejan el
proceso de path creation. Empresas con una estrategia basada en la innovación logran
acumular las competencias necesarias para aprovechar las oportunidades dadas por el
entorno y evitar sus restricciones y amenazas. Esto explica por qué algunas firmas están
mejor preparadas para lidiar con entornos inestables.
- v -
Acknowledgements
This thesis is the result of team work, of long hours of discussion, analysis, and research,
and for that I sincerely thank those who joined me in the process. To Guille, who stood by
my side unconditionally throughout the long hours in Buenos Aires, the short minutes of
each long-distance phone call and joined me throughout the dark Danish winter. I could
not have done it without you.
To my family. To my mom and dad, who taught me to fight for what I want and convinced
me that “I can.” To Lore, for her great advice that reinforced this. To Oski, because he
taught me to study listening to rock. To my grandmother, for all her good wishes. To my
uncle Jorge and pastries. To my dearest friend Sele, who always forgives my forgetfulness.
To Nina for understanding the silence. To my nana.
I want to thank my supervisors and my colleagues, because they were always there to read,
to comment, and to listen to me in moments of catharsis. To Gustavo Lugones, who
taught me to overcome the blank page. To Gabriel Yoguel, for his insistence and
understanding. To José Borello for his advice. To Bengt-Åke Lundvall, because he trusted
me and gave me a unique opportunity. To Björn Johnson, for his patience and willingness
to deal with my ideas. To Bernardo Kosacoff, for his willingness to help when needed. To
my colleagues from the IdeI: Veronica, Florencia, Mariano, Sonia, Dario, Vlad, Analía, and
Sebastián, for their support and comments. To my fellows Roman, Daniel, and Eunkyung,
for making me feel I was not alone. To Fernando Porta, Jorge Katz, Juan Carlos Hallak,
Birgitte Gregersen, and the IKE group members: Jacob, Bram, Max, Alex, Gert,
Annemarie, and Kristian, for the time they took to read my progress. To Gro Villumsen
and Juan Luis Merega because they didn’t gave up. To Dorte and Jeanette for all the help
during my stay at the AAU. To Victoria for proofreading the English text.
Again, to all of you, thanks.
Diana
Agradecimientos
Esta tesis es fruto del trabajo colectivo, de largas horas de discusión, de análisis y de
investigación, y por ello quiero agradecer profundamente quienes me acompañaron en el
Acknowledgements
- vi -
proceso. A Guille, quien con su apoyo incondicional estuvo a mi lado en las largas horas en
Buenos Aires, en los breves minutos a la distancia y en los oscuros días del invierno danés.
No podría haberlo hecho sin vos.
A mi familia. A mamá y papá que me enseñaron a luchar por lo que quiero y me
convencieron del “yo puedo”. A Lore por sus grandes consejos, que lo reforzaron. A Oski,
porque me enseñó a estudiar escuchando rock. A mi madrina, por sus saludos y buenos
augurios. A mi tío Jorge y sus facturas vespertinas. A mi amiga del alma Sele, quien día a día
perdona mis olvidos. A Nina, por entender los silencios. A mi nana.
Quiero agradecer también a mis supervisores y a mis colegas, porque siempre estuvieron
presentes para leer, para comentar y para acompañarme en los días de catarsis. A Gustavo
Lugones, que me enseñó a vencer la hoja en blanco. A Gabriel Yoguel por su insistencia y
comprensión. A José Borello por sus consejos. A Bengt-Åke Lundvall porque confió en mi
y me brindó una oportunidad única. A Björn Johnson, por la paciencia y la predisposición a
lidiar con mis ideas. A Bernardo Kosacoff, porque siempre estuvo dispuesto a ayudarme en
lo que necesitara. A mis colegas del IdeI: Verónica, Florencia, Mariano, Sonia, Darío, Vlad,
Analía, y Sebastián, por su apoyo y disposición. A mis compañeros Roman, Daniel,
Eunkyung y Juan, porque me hicieron sentir que no estaba sola. A Fernando Porta, Jorge
Katz, Juan Carlos Hallak, Birgitte Gregersen, y los miembros del grupo IKE: Jacob, Bram,
Max, Alex, Gert, Annemarie, y Kristian, por el tiempo que dedicaron a leer mis avances. A
Gro Villumsen y Juan Luis Mérega porque no se dieron por vencidos. A Dorte y Jeanette
por su ayuda durante mi estadía en la AAU. A Victoria por sus correcciones.
Nuevamente, a todos, gracias.
Diana
Table of contents
- vii -
Table of contents
Table of contents ........................................................................................................................................ vii
List of abbreviations .................................................................................................................................... xi
List of figures, graphs, and tables ............................................................................................................. xii
Introduction ............................................................................................................................ 1
I. Motivation .................................................................................................................................................. 1
II. Research project and structure of the thesis ....................................................................................... 1
III. Research question and objectives ........................................................................................................ 8
IV. Methodological questions and hypothesis ......................................................................................... 8
SECTION I - THE NATIONAL INNOVATION SYSTEM APPROACH ............ 10
Introduction to Section I ........................................................................................................................... 10
Chapter 1: From a “focusing device” to a matter of public policy ........................................ 12
Introduction ................................................................................................................................................. 12
1.1. Theoretical background: innovation economics ............................................................................ 13
1.1.1. The capitalist dynamic ................................................................................................................. 14
1.1.2. Defining innovation .................................................................................................................... 16
1.1.2.1. Innovation at the macro level ................................................................................................. 16
1.1.2.2. Innovation at the meso level ................................................................................................... 18
1.1.2.3. Innovation at the micro level .................................................................................................. 20
1.2. The Innovation System approach ..................................................................................................... 22
1.2.1. The four founding fathers: basic definitions ........................................................................... 22
1.2.2. Building blocks for an IS: dimensions of analysis .................................................................. 26
1.2.2.1. Organizations and institutions ................................................................................................ 27
1.2.2.2. The firm ..................................................................................................................................... 30
1.2.2.3 System and environment .......................................................................................................... 33
1.2.2.4. Interactions within the IS: cooperation and interdependence .......................................... 36
1.2.2.5. Knowledge and competence building ................................................................................... 38
1.3. The IS approach and unstable environments ................................................................................. 39
1.3.1. Development as a relative concept ........................................................................................... 39
1.3.1.1. Development and instability ................................................................................................... 42
1.3.2. Development and the IS approach ........................................................................................... 45
1.3.3. The Latin American view ........................................................................................................... 48
1.3.3.1. A specific approach vs. a specific approximation to development .................................. 49
1.3.3.2. Causes and consequences of development........................................................................... 50
1.3.3.3. Theory and practice .................................................................................................................. 51
1.4. Preliminary conclusions ...................................................................................................................... 53
Chapter 2: A historical analysis of the process of competence building in an unstable
system .................................................................................................................................... 56
Introduction ................................................................................................................................................. 56
2.1. Methodological discussion ................................................................................................................. 57
2.2. An overview of the Argentinean NIS .............................................................................................. 60
2.2.1. The agro-export model (1880–1930) ........................................................................................ 60
2.2.2. Industrialization by import substitution (ISI) (1930–1976) .................................................. 63
Table of contents
- viii -
2.2.2.1. More than an industrialization policy .................................................................................... 63
2.2.3. The Washington Consensus (1976-2002) ................................................................................ 65
2.2.3.1. The return to liberalism ........................................................................................................... 65
2.2.3.2. The 1990s and structural reforms .......................................................................................... 66
2.2.3.3. Science and technology institutions during the Convertibility Plan ................................. 68
2.2.3.4. The productive structure: specialization and static comparative advantages .................. 69
2.2.3.6. The labor market and domestic demand .............................................................................. 70
2.2.4. “Argentina for everyone”: recovery, growth, and development .......................................... 71
2.3. Innovation and competence building: structural characteristics .................................................. 73
2.3.1. Industrialization versus comparative advantages .................................................................... 75
2.3.2. Growth versus recession ............................................................................................................ 76
2.3.3. Public versus private goods ........................................................................................................ 78
2.3.4. Virtuous behavior versus rent-seeking behavior .................................................................... 80
2.4. Preliminary conclusions ...................................................................................................................... 81
SECTION II - THE INNOVATIVE STRATEGIES .............................................. 83
Introduction to Section II ......................................................................................................................... 83
Chapter 3: The strategy-based approach to firm behavior .................................................... 85
Introduction ................................................................................................................................................. 85
3.1. Capabilities, resources and firm behavior ........................................................................................ 86
3.1.1. Evolutionary approaches to firm behavior .............................................................................. 86
3.1.1.1. The capability-based approach ............................................................................................... 87
3.1.1.2. The resource-based approach ................................................................................................. 90
3.1.1.3. The knowledge-based approach ............................................................................................. 91
3.1.2. Consensus and limitations of the approaches ......................................................................... 93
3.2. Strategic decisions and innovation dynamics .................................................................................. 97
3.2.1. Innovative strategies and observable structures ..................................................................... 97
3.2.1.1. The evolutionary concept of the firm’s innovative strategy .............................................. 98
3.2.1.2. Freeman’s innovative strategies ............................................................................................100
3.2.1.3. Pavitt’s sectoral taxonomy .....................................................................................................102
3.2.2. The static perspective of the strategic process ......................................................................103
3.3. The strategy-based approach: towards a dynamic theory of the innovative firm ...................104
3.3.1. Learning processes and interactions with the environment ...............................................106
3.3.2. Innovative strategies and competence building ....................................................................109
3.3.2.1. Persistence of innovation ......................................................................................................110
3.3.3. The new process of strategic decision and the impact of the environment .....................112
3.3.1. Environmental shocks and resilience .....................................................................................114
3.4. Preliminary conclusions ....................................................................................................................116
Chapter 4: Towards a conceptual framework for innovative strategies ............................... 118
Introduction ...............................................................................................................................................118
4.1. Literature review: strategies, results, and impacts ........................................................................120
4.1.1. Empirical evidence on innovative strategies .........................................................................121
4.1.1.1. Discussion of the evidence on innovative strategies ........................................................124
4.1.2. Innovation results: persistence of innovation .......................................................................127
4.1.2.1. Some gray areas in studies about persistence .....................................................................129
4.1.3. Innovative impacts: productivity and growth .......................................................................133
4.1.3.1. Nelson and Winter’s relationship between innovation and performance .....................134
4.1.3.2. Crépon, Duguet, and Mairesse’s model of results and productivity ..............................136
- ix -
4.1.4. Environmental determinants: structural variables ................................................................138
4.2. Conceptual approach and discussion of the hypotheses ............................................................140
4.2.1. Innovative strategies and innovative structures ....................................................................141
4.2.2. Innovative strategies and performance ..................................................................................145
4.3. Preliminary conclusions ....................................................................................................................146
SECTION III: INNOVATIVE STRATEGIES IN THE ARGENTINEAN
MANUFACTURING SECTOR .............................................................................. 148
Introduction to Section III ......................................................................................................................148
Chapter 5: The dataset and methodology ........................................................................... 150
Introduction ...............................................................................................................................................150
5.1. The dataset..........................................................................................................................................150
5.1.1. ENIT - Argentinean Innovation Surveys ..............................................................................150
5.1.2. The panel and the population of manufacturing firms ........................................................151
5.1.2.1. Size: sales and employment ...................................................................................................151
5.1.2.2. Activity sector .........................................................................................................................152
5.1.2.3. Capital ownership of the firm ...............................................................................................153
5.1.2.4. Innovation investments and results .....................................................................................154
5.1.3. The panel: cross-relationships and distribution ....................................................................155
5.1.4. Selection bias and representativeness .....................................................................................157
5.2. The selected variables: definitions and treatment ........................................................................160
5.2.1. Segmentation of periods and general data treatment ...........................................................160
5.2.2. The innovative strategies ..........................................................................................................160
5.2.3. Innovation investments ............................................................................................................163
5.2.4. Innovation results ......................................................................................................................165
5.2.5. Qualified human resources.......................................................................................................166
5.2.6. Linkages with the National Innovation System ....................................................................167
5.2.7. Financial resources ....................................................................................................................168
5.2.8. Performance: productivity, employment, and exports ........................................................169
5.2.9. Environmental characteristics: size, sector, capital ownership, and sub-periods ............170
5.3. Integration of the conceptual and methodological approaches .................................................171
5.4. Preliminary conclusions ....................................................................................................................176
Chapter 6: Regularities and changes in firms’ strategies .................................................... 178
Introduction ...............................................................................................................................................178
6.1. Hypotheses and Methodology .........................................................................................................178
6.1.1. Hypotheses .................................................................................................................................178
6.1.2. The model ...................................................................................................................................180
6.1.2.1. The descriptive approach ......................................................................................................182
6.1.2.2. The cluster analysis .................................................................................................................182
6.2. Results from the descriptive analysis ..............................................................................................184
6.2.1. The innovative strategies ..........................................................................................................184
6.2.2. The strategies and the innovative structure ...........................................................................187
6.3. Results from the cluster analysis .....................................................................................................191
6.4. Discussion of the findings ...............................................................................................................196
6.5. Preliminary conclusions ....................................................................................................................200
Chapter 7: Innovative strategies, innovation persistence, and firm performance .............. 203
Introduction ...............................................................................................................................................203
7.1. Hypotheses and methodology .........................................................................................................204
- x -
7.1.1. Hypotheses .................................................................................................................................204
7.1.2. The model ...................................................................................................................................206
7.1.2.1. General overview and the inclusion of strategies ..............................................................206
7.1.2.2. Innovation persistence ...........................................................................................................208
7.1.2.3. Strategies, innovation, and performance .............................................................................210
7.2. Results from the descriptive analysis ..............................................................................................212
7.3. Estimation of the model...................................................................................................................216
7.3.1. Innovation persistence ..............................................................................................................216
7.3.2. Productivity and innovative strategies ....................................................................................220
7.3.3. Employment growth and innovative strategies ....................................................................222
7.3.4. Exports and innovative strategies ...........................................................................................224
7.4. Discussion of the findings ...............................................................................................................226
7.5. Preliminary conclusions ....................................................................................................................230
SECTION IV: CONCLUSIONS .............................................................................232
8.1. Research questions and findings .....................................................................................................232
8.2. Theoretical implications ...................................................................................................................234
8.2.1. The NIS and micro-heterogeneity ..........................................................................................234
8.2.2. The NIS and innovative firm ...................................................................................................236
8.2.3. Innovative strategies and unstable environments .................................................................237
8.3. Methodological implications ............................................................................................................240
8.3.1. Innovative strategies and micro-heterogeneity .....................................................................241
8.3.2. Results about the persistence of innovation ..........................................................................242
8.3.3. Innovation process, results, and impacts ...............................................................................244
8.4. Practical implications ........................................................................................................................245
8.4.1. Application to other countries .................................................................................................245
8.4.2. Policy recommendations ..........................................................................................................246
8.5. Limitations and future lines of work ..............................................................................................248
Bibliography ........................................................................................................................ 250
Appendix ............................................................................................................................. 262
Appendix 1: The panel and the definition of the variables ................................................................262
1.1. Methodology and samples of the National Innovation Surveys ...........................................262
1.2: Definition and matching of the variables ..................................................................................262
1.3: The innovation balance index .....................................................................................................265
1.4: Capital ownership – comparison of selected variables ...........................................................266
Appendix 2: The innovative strategies ..................................................................................................267
2.1. Non-parametric and multivariate statistics – Methodology ...................................................267
2.2. Non-parametric estimations - Results .......................................................................................269
2.3. Multivariate analysis of variances - Marginal effects ...............................................................273
2.4. Discriminant analysis ....................................................................................................................274
Appendix 3: Innovative strategies, persistence and performance .....................................................277
3.1. Non-parametric estimations ........................................................................................................277
3.2. Persistence of innovation (probit and zero inflated models) .................................................281
3.3. Performance and strategies (OLS models) ...............................................................................282
- xi -
List of abbreviations
AEM Agro-export model ALTEC Latin American association for technological management (from Spanish: Asociación
Latinoamericana de la Gestión Tecnológica) ANOVA Analysis of variance ANPCyT National Agency for the Promotion of Science and Technology (from Spanish:
Agencia nacional de promoción científica y tecnologica) BDDE Entrepreneurial database (from Spanish: Base de datos de desempeño empresarial) BRICS Brazil, Russia, India, China, South Africa CEP Center of production studies (from Spanish: Centro de estudios de la produducción) CIS Community innovation survey CMD Crépon, Duguet, and Mairesse’s model CNEA National Commission of Atomic Energy (from Spanish: Comisión nacional de
energía atómica) CONICET National Council of Scientific and Technical Research (from Spanish: Consejo
nacional de investigaciones científicas y técnicas) DF Discriminant function DUI Doing, using, and interacting ECLAC Economic commission for Latin America and the Caribbean (in Spanish: CEPAL) ENIT Argentinean innovation surveys (from Spanish Encuesta Nacional sobre innovación y
conducta tecnológica) EU European Union FDI Foreign direct investment GACTEC Science and Technology Council (from Spanish: Gabinete científico tecnológico) GDP Gross domestic product GERD Gross expenditure on R&D GLOBELICS Global Network for Economics of Learning, Innovation
and Competence Building Systems IMF International monetary fund INDEC Argentina’s statistitcal institute (from Spanish: Instituto nacional de estadísticas y
censos) INTA National Institute of Agricultural Technology (from Spanish: Instituto nacional de
technología agropecuaria) INTI National Institute of Industrial Technology (from Spanish: Instituto nacional de
tecnología industrial) IS Innovation system IKE Knowledge and Economic Dynamics group ISI Industrialization by import sustitution (from Spanish: industrialización por
sustitución de importaciones) ISIC International Standard Industrial Classification MANOVA Multivariate analysis of variances MERCOSUR Common market of the south (from Spanish: mercado común del sur) MINCyT Argentinean S&T Ministry (from Spanish: Ministerio de ciencia, tecnología e
innovación productiva) NACE Statistical classification of economic activities in the European Community NIS National innovation system OECD Organization for economic cooperation and development OLS Ordinary least squares PPI Producer price index R&D Research and development RE Random effects RICYT Iberoamerican Network of Science and Technology Indicators (from Spanish: Red
iberoamericana de indicadores de ciencia y tecnología) S&T Science and technology
List of abbreviations, figures, tables and graphs
- xii -
SID System of innovation for development SPU National secretry of University policies(from Spanish: Secretaría de políticas
universitarias) STI Science, technology and innovation UIS-UNESCO UNESCO institute for statistics WB World Bank
List of figures, graphs, and tables
Figures
Figure 1.1: Main theoretical contributions to the NIS approach ........................................................ 14
Figure 1.2: Key building blocks of the IS approach .............................................................................. 27
Figure 2.1: Conceptual scheme of the National Innovation System .................................................. 59
Figure 2.2: Summary of the history of the Argentinean innovation system ...................................... 74
Figure 4.1.a: The strategy-based approach—The innovative strategy ..............................................141
Figure 4.1.b: The strategy-based approach—The non-innovative strategy .....................................141
Graphs
Graph 1.1: Gross domestic product (GDP) and income distribution (GINI)- 2008 ...................... 41
Graph 1.2: Gross domestic product (GDP) and expenditure on R&D (GERD) - 2008 .............. 41
Graph 1.3: Gross domestic product (1980=1)—Selected countries (1980–2006) ........................... 43
Graph 1.4: Argentina and Denmark (1980-2006)—Selected indicators (1983=1) .......................... 43
Graph 6.1: Strategies and discriminant functions (DF) ......................................................................193
Graph 6.2.a: Discriminant Function 1 versus Discriminant Function 2 .........................................194
Graph 6.2.b: Discriminant Function 1 vs. Discriminant Function 3 ...............................................195
Graph 6.3: Innovative strategies and innovative structure.................................................................197
Graph 6.4: Composition of the strategies—Selected sectors (% of firms) .....................................198
Graph 7.1: Average annual growth rate 1998-2006—Productivity, growth, and exports ............229
Tables
Table 5.1: Size distribution—Panel and ENIT (% of firms) .............................................................152
Table 5.2: Sectoral distribution—Panel and ENIT (% of firms) ......................................................153
Table 5.3: Capital Ownership—Panel and ENIT (% of firms).........................................................154
Table 5.4: Innovative dynamics—Panel and ENIT (selected indicators) ........................................154
Table 5.5: Size evolution of the panel (breakdown by quartiles of employment) ..........................156
Table 5.6: Sectoral distribution by quartiles and capital ownership (Panel=100 % of firms) ......157
Table 5.7: Capital ownership and size (Panel=100 % of firms) ........................................................157
Table 5.8: Representativeness of the panel (selected indicators) .......................................................159
Table 5.9: Combinations of innovative strategies ................................................................................162
Table 5.10: Innovation expenditure categories ....................................................................................164
Table 5.11: Summary of the dimensions and variables .......................................................................172
Table 5.12: Innovative structure—Correlation matrix .......................................................................173
Table 5.13: Innovation results and performance—Correlation matrix ............................................175
Table 6.1: Combinations of innovative strategies ................................................................................179
Table 6.2: Summary of variables .............................................................................................................181
List of abbreviations, figures, tables and graphs
- xiii -
Table 6.3: Innovative strategies—Number of firms (Pearson’s std. residuals) ...............................185
Table 6.4: Frequency of expenditure—Number of years—Mean (Std. dv.) ...................................186
Table 6.5: Innovation intensity (Iexp) and balance (IB)–Mean (Std. dv.)........................................187
Table 6.6: Distribution of the innovation expenditure .......................................................................189
Table 6.7: Qualified human resources (QHR) and linkages (Link) ...................................................189
Table 6.8: Linkages with organizations from the NIS ........................................................................190
Table 6.9: Access to external resources (ER)—% of firms (std. residuals) .....................................191
Table 6.10: MANOVA.............................................................................................................................192
Table 6.11: Canonical structure (coefficients of the discriminant functions) .................................192
Table 6.12: Strategies and discriminant functions–Mean values (std. dv.) ......................................195
Table 6.13: Cluster analysis - Estimated and observed cluster membership ...................................196
Table 7.1: Innovation persistence: summary of the dimensions and variables ...............................210
Table 7.2: Performance: summary of the dimensions and variables ................................................212
Table 7.3: Innovation persistence and strategies——% of firms (std. residuals) ..........................213
Table 7.4: Productivity and innovative strategies (ARS 000)—Mean (std. dv.) ..............................214
Table 7.5: Employment—number of people—Mean (std. dv.) ........................................................215
Table 7.6: Exports—% of firms and intensity .....................................................................................216
Table 7.7: Random effects dynamic probit model—Dep. variable: Innovations (Inno) ..............218
Table 7.8: Marginal effects - RE dynamic probit—Dep. variable: Innovations (Inno).................219
Table 7.9: Random effects dynamic regression model—Dep. var.: Productivity (Pd) ..................222
Table 7.10: Random effects dynamic regression model—Dep. variable: Growth (Gr) ................224
Table 7.11: Random effects dynamic probit model—Dep. variable: Exports (Xpo) ....................225
- 1 -
Introduction
I. Motivation
This thesis is about innovation, instability, and micro-heterogeneity. The theoretical
motivation behind the research focuses on the Argentinean case and arises from two sets
of innovation studies about Argentina carried out in recent decades. One set of analyses
explains that the unstable Argentinean environment has historically worked against
innovation (Dutrenit and Katz, 2005; Herrera, 1971; Katz and Kosacoff, 1998; Kosacoff
and Ramos, 2006; López, 2003; Anlló et al., 2007; Porta, 2003, 2006; Prebisch, 1950).
Another set of studies shows the existence of some Argentinean firms that have
successfully innovated and grown (Borello et al., 2005; Chudnovsky et al., 2004; Erbes et
al., 2004; Katz, 1976; Kosacoff, 1998; Lugones et al., 2008). In many cases, the same
scholars are behind both sets of research works, but none have provided a general
explanation for why some firms managed to cope with instability and gain competitiveness
based on innovation. This is precisely the research question that guides this thesis: why are
some firms better prepared to cope with unstable environments than others?
During the last century, the main constant in Argentina’s economic history has been the
recurrent cycles of recession, growth, and changes in the rules of the game (Bisang, 1993,
1995; Diamand, 1983; Hirschman, 1968; Katz and Kosacoff, 1998). Within this context of
instability, the period 1998–2006 constitutes a good example of the unstable environment
the Argentinean firms have had to deal with. From a macroeconomic point of view, this
period entailed a deep recession between 1998 and 2001 (a 18.3% drop in GDP), a change
in the economic model plus a quick recovery between 2002 and 2004 (a 18.6% increase in
GDP), and high rates of growth during 2005 and 2006 (an average annual rate of around
8%). From an institutional perspective, this was accompanied by the appointment of five
presidents in less than one month, the confiscation of fixed-term deposits, recurrent
demonstrations and strikes, and the dollarization of economic activity. A new local
vocabulary was coined to describe these changes: pesificación (the compulsory conversion of
savings and loans from dollars into pesos), piquetes (blocking streets as a way of protesting),
cacerolazos (mass banging of pots and pans in public places to demand changes in public
policies), and corralito (restrictions to withdrawing money from bank accounts).
Introduction
- 1 -
During this period, firms followed different competitive paths in an attempt to deal with
the changing environment, some of which were based on innovation and technological
change, while others were not based on innovation at all (Anlló et al., 2007; López and
Arza, 2008; Suárez, 2009). Both groups contain firms that grew and survived the
macroeconomic fluctuations. This provides evidence regarding the existence of micro-
heterogeneity (in the sense of Bottazzi et al., 2010; Dosi, et al., 2010; Nelson, 1991a, b;
Penrose, 1959; Salter, 1960) and the importance of the concept when accounting for the
Argentinean situation. This is the practical motivation of this research: to answer why
“good” firms can be found in the “bad” Argentinean environment.
II. Research project and structure of the thesis
ii.1. Section I - The National Innovation System approach
The search for answers starts with the selection of the theoretical framework to tackle the
relationship between firms and the environment (Chapter 1). The National Innovation
System (NIS) approach was selected, due to its ability to account for the existence of
different firms —the variety of the system— and the way that the specific environment
where firms operate, the characteristics of the rest of the organizations, and the interactions
between them —systemic relationships— shape the intensity and direction of innovation
and technological change (Freeman, 1995; Lundvall, 1992; Nelson, 1993).
Just as the firm is more than the sum of its human resources and assets (Nelson and
Winter, 1982; Penrose, 1959), the system is more than the sum of its sectors and
organizations (Lundvall, 1992; Lundvall et al., 2009a). The selection of this approach and
its application to the Argentinean case seeks to provide an analysis capable of
complementing the partial explanations of sectoral/local analysis and the anecdotal nature
of case studies (Breschi and Malerba, 1997, Cassiolato et al., 2000; Malerba and Nelson,
2007, among many others).
At the national level, in terms of the NIS approach, there are two types of explanation
about the relationship between innovation at the firm level and unstable environments.
One group is made up of studies that mainly derive from cross-country comparisons,
which conclude that underdevelopment is a matter of distances between developed and
developing nations for a set of key innovative dimensions (Alburquerque, 1999; CEPAL,
2008; Cimoli, 2005; Godinho et al., 2004; Nelson and Dahlman, 1991; OECD, 2007,
among others). In this sense, they explain the distances between countries in terms of a
Introduction
- 2 -
systemic approach which includes different forms of knowledge creation (research,
technology transfer, acquisition of capital goods, etc.) but also skilled human resources,
high-tech exports, types of interactions, etc. These studies sustain that there is a
multidimensional gap between Argentinean firms and those from developed countries, and
that the way to close the gap is through investments in innovation.
The second type of explanation is mainly found among scholars from Argentina and Latin
America, who argue that underdevelopment is the result of historical economic and
institutional instability and a productive structure that is biased towards low-tech and low
income-elasticity sectors (Dutrenit and Katz, 2005; Katz, 2007; Lugones and Suárez, 2010;
Porta, 2006, Bisang, 1995; among others). In this case, the low-profile innovative dynamics
among Argentinean firms have to be overcome by means of structural change together
with a set of policies aimed at providing a more stable and less volatile macroeconomic
environment.
Both types are valid and well-proven explanations of the low innovative performance of
Argentinean firms. There is no doubt that there is a multidimensional gap between
developed countries and Argentina, but it is the result of historical processes of
disarticulation between the productive sector, knowledge infrastructure, and public policies,
together recurrent cycles of recession followed by unequal growth (Chapter 2). The
emphasis on distances within the first explanation may lead to forgetting that they are the
result (and not the cause) of the dynamics of the system. In turn, the emphasis on the
structural dimension in the second explanation works against the possibility of identifying
and characterizing firms that were actually capable of escaping environmental constraints.
Neither of these analyses can explain why some firms deal better with this unstable
environment by showing a high-profile innovative strategy and managing to grow. In this
sense, the existent analysis based on the NIS approach cannot provide a satisfactory answer
to these questions. In the first case, the problem seems to be the static perspective on the
dynamics of the firm’s innovation process; in the second case, it seems to lie in the lack of
attention given to the heterogeneity of the system.
In short, twenty years after the publication of Lundvall’s (1992) and Nelson’s (1993) books,
the NIS approach seems to have become a somewhat static way of looking at a complex
reality, in which learning and competence building have been reduced to inputs and
outputs of the innovation process. Somehow, the “black box” is back: the innovative
dynamic at the firm level has been controlled by size- and sector-based classifications, the
Introduction
- 3 -
feedback and accumulation processes have been turned into R&D expenditure and patents;
the systemic nature of innovation has been narrowed to the frequency of cooperation
agreements; and historical characteristics have been interpreted as barriers to development.
However, two decades of studies have also contributed with a body of knowledge about
the Latin American case that has not been yet fully integrated into the theory (Arocena and
Sutz, 1999; Cassiolato and Lastre, 2002, 2008; Dutrenit, 2009; Dutrenit et al., 2010, to name
a few).
Discussing and analyzing the approach and how it helps the Argentinean reality to be
understood is the objective of Section I. Chapter 1 reviews the foundations and the recent
literature on NISs by focusing mainly on two dimensions: firstly, the potentialities and
limitations of this approach when explaining the innovative dynamics of an unstable
country; secondly, how the micro and the macro dimensions of the approach interact and
generate feedbacks which can improve or block the development of the system. Chapter 2
presents the Argentinean case and applies the approach in order to identify in which
respects it can answer the questions posed and in which respects it cannot. It also shows
why Argentina is characterized as an unstable innovation system.
The conclusions from Section I are that there is a missing element in the theoretical
foundations of the approach, as the interactive relationship between the NIS and firm
behavior is not properly accounted for. The concept of variety does not sufficiently
account for why some firms are better prepared to deal with the unstable environment than
others, while the static characteristics of the approach have undermined the capacity to
explain how firms deal with change. In the quest for answers to our questions, Section II
continues by zooming in on firms’ levels, relating innovation strategies to the systemic
perspective.
ii.2. Section II - The innovative strategies
Section II proposes an analysis of micro-dynamics and how firms learn to compete,
survive, and grow within national systems where the theory predicts mostly short-term
opportunistic behavior. To do this, the evolutionary theory of the firm will be examined
(Freeman, 1974; Freeman and Soete, 1997; Jensen et al., 2007; Langlois and Robertson,
1995; Lall, 1992, 2001, Lundvall and Lam, 2007; Metcalfe, 2002; Nelson, 1991a, b, 1994;
Nelson and Winter, 1982; Pavitt, 1984; Penrose, 1959; Teece and Pisano, 1994) in order to
analyze how the literature deals with the existence of micro-heterogeneity and how it can
explain the existence of firms that learnt to deal with unstable environments, not
Introduction
- 4 -
necessarily by innovating. Theoretical approaches to firm behavior are found to explain the
existence of micro-heterogeneity associated with differences in the firm’s resources and
capabilities (Lundvall and Lam, 2007; Nelson, 1991a, b; Penrose, 1959), the impact of the
selection process (Bottazzi et al., 2010; Dosi et al., 2010; Nelson and Winter, 1982), and the
impact of path dependence (David, 2007; Antonelli, 1997). However, none of the
explanations account for the relationship between micro-heterogeneity and the NIS —
since this relationship is greater than just the selection process— nor the way a firm
manages to escape from the constraints it can create.
To explain this relationship and to shed light on how firms deal with an unstable
environment, this section proposes the concept of strategy (Chapter 3). The strategy is
the set of decisions regarding how to tackle the selection process. It arises from the firm’s
interactions with the environment (and the NIS), and its resources and capabilities (path
dependence). Since firms can learn, they are able to re-interpret the environment and adjust
how they allocate and exploit their resources, that is to say, they adjust their strategy. I will
show that the continuum between the process of strategic decisions, implementations, and
adjustments is key to understanding the sources of path creation (a micro-interpretation of
Garud et al., 2010 and Lovio et al., 2011) and how learning and accumulation triggers
different results in terms of the firm’s performance within a specific environment.
The main conclusion from this section is that firms have different capacities for learning
how to deal with the environment. Given that firms learn, some of them will accumulate
capabilities for taking advantage of opportunities or avoiding environmental risks. In the
case of unstable environments, some firms will be able to escape the structural
determinants and follow a high-profile innovative strategy. This dynamic process of
learning and reacting cannot be studied by means of input and output measures but
requires the systemic nature of the innovation process to be taken into account. This
study’s theoretical analysis (Chapter 3) and review of the empirical literature (Chapter 4)
suggests that how firms innovate—regardless of how many innovations they have
achieved—is what explains the different trajectories, the different performances, and the
continuity of “better” behavior over time. From a practical point of view, the analysis
herein suggests, on the one hand, that the continuity of innovative investment is a
useful element for identifying patterns of micro-heterogeneity and understand the
impact of the innovative strategy. On the other hand, it suggests that, given the changes
Introduction
- 5 -
in the Argentinean environment, some firms may have changed their strategy and
this is a key element to understanding how micro-heterogeneity changes over time.
This section reveals that the gap in the literature is the lack of dynamism when it comes to
firm behavior. Although the theoretical argumentations provide a clear view about the
dynamic process between the firm’s behavior, the competition in the market, and the
resultant adjustments inside the firm, empirical analysis cannot explain how a firm reacts to
the environment in such a way that it radically changes its behavior (Antonelli, 1997;
Clausen et al., 2011; Leiponen and Drejer, 2007; Raymond et al., 2010; Srholec and
Verspagen, 2012, among others). The path dependence view attached to this literature
works against its ability to explain how firms’ decisions lead to “path creation,” which is the
expected outcome of an organization that is able to learn (in the sense of Cohen and
Levinthal, 1990; Jensen et al., 2007; Lundvall and Lam, 2007). Within this static view, the
innovation process, the innovation results, and the firm’s performance become a linear
process linking different combinations of inputs to different levels of outputs. Such texts
make mention of the feedbacks and signals from the environment, but do not analyze how
they alter the firm’s trajectory. In other words, instead of a black box, the literature now
presents a heterogeneous structure of boxes, but no less black than before.
Consequently, the next step in the search for answers is to empirically study the existence
of different strategies and the different impacts they have on the firm’s performance. This
is the focus of Section III, where I apply the strategy-based approach to the Argentinean
case and propose a methodology to dynamically analyze the innovation process and how it
impacts the firm’s performance, allowing for the possibility of firms changing their
strategies.
ii.3. Section III - Innovative strategies in the Argentinean manufacturing sector
The objective of Section III (Chapters 5 to 7) is to apply the conceptual framework linking
innovative strategies to unstable innovation systems. The conceptualization of the
innovative strategies will be tested for a group 800 Argentinean manufacturing firms.
The analysis is based on information from the Argentinean National Innovation
Survey (Encuesta Nacional sobre innovación y conducta tecnológica, hereafter
ENIT) for the period 1998–2006 (INDEC, 2003, 2006, 2008, 2010). Since these firms
have been operating in a changing environment, the objective is to identify and characterize
continuities and discontinuities in the firm’s behavior and, especially, changes in this
behavior over time.
Introduction
- 6 -
The model proposes a dynamic approach to the firm in which the innovation process, the
innovation results, and the firm’s performance interact, subject to the existence of path
dependence and accumulation processes, but allowing for the possibility of a change in the
firm’s behavior. The methodology will be based, firstly, on a cluster analysis (Klecka,
1980; Lachenbruch, 1975; Tu and Han, 1982) in which changes between categories of firms
are allowed, and, secondly, on probabilistic and regressive estimations (Crépon et al.,
1998; Peters, 2009; Raymond et al., 2010; Wooldridge, 2005) in order to measure the
strength and the sign of the relationship between the strategies, the innovation results, and
the firm’s performance (Chapter 5).
In this section, the contribution of this study is the proposal of a dynamic model capable of
explaining how firms move between strategies or, in terms of the literature, between
different innovative behaviors (Chapter 6). The literature on innovative behavior accounts
for different innovative statuses and whether the distribution of firms between statuses
changes over time —more or less firms investing in innovation, more or less firms
achieving innovations, firms with different combinations of innovation activities, etc.
(Arundel and Hollanders, 2005; Clausen et al., 2011; de Jong and Marsili, 2006; Freeman
and Soete, 1997; Huergo and Moreno, 2011; Ito and Lechevalier, 2010; Pavitt, 1984;
Srholec and Verspagen, 2012; Yurtseven and Tandoğan, 2012). The results show the
importance of the way firms change from one situation to another (that is, how they
change their strategy). Moreover, methodologically speaking, the results suggest that the
patterns of change in the frequency of innovation investments have significant predictive
power in terms of the evolution of the level and characteristics of firms’ innovative
behavior and capabilities.
Another contribution to the literature has to do with the analysis of the link between the
innovative strategy and the performance of the firm (chapter 7). In this case, the literature
sustains that there is a positive relationship between past and present innovations, which is
summarized using the concept of persistence (Antonelli, 1997; Geroski et al., 1997;
Malerba et al., 1997; Raymond et al., 2010). The results of this analysis challenges that
concept and show that, within unstable environments, the correlation between past and
present results is explained more by the firm’s innovative behavior than by the results it has
achieved in the past. I will look also at how the relationship between innovation activities,
innovation results and performance is assumed to be positive (and linear) (Crépon et al.,
1998; Coad and Rao, 2011; Molina-Domene and Pietrobelli, 2012) and show that
Introduction
- 7 -
depending on the specific definition of performance (productivity, growth,
competitiveness) innovation may or may not have an impact at all, or may even lead to a
negative relationship, depending on the firm’s innovative strategy. When it comes to
economic performance, the results suggest that whether firms invest continuously in
innovation and capabilities is more important than how much they have invested or how
frequently they have successfully innovated.
The preliminary conclusions from this section are that the concept of the innovative
strategy helps the dynamic process of innovation and reactions to changes in the
environment to be better characterized, which in turn help us understand why high-profile
innovative firms with good economic performances can be found within a context of
macroeconomic and institutional instability. In other words, it allows to answer why
“good” firms can be found in the unstable Argentinean environment.
ii.4. Section IV - Conclusions
The last section (Chapter 8) discusses the conclusions of the thesis with respect to the
research question and objectives. My analysis leads to the preliminary conclusion that
micro-heterogeneity is endogenous to the dynamics of the NIS, where firms will
adjust their strategies differently according to changes in the environment. This study
shows that Argentinean firms facing the same unstable environment react differently in
terms of their innovative strategy and this impacts the aggregate innovation process of the
system and the interactions between the firms and the rest of the organizations. This
provides evidence regarding the importance of studying the firm’s behavior from a
dynamic perspective, allowing not only for the firm to modify the way it seeks innovations
but the possibility of not seeking innovations at all. My findings suggest that the firm’s
learning capabilities and innovation investments are key factors in understanding why some
firms are better prepared to deal with unstable environments, not because they are more
likely to successfully innovate but because they have dealt better with the process of path
creation. Firms with an innovation-based strategy have accumulated the competencies
required to take advantage of environmental opportunities and avoid environmental
constraints. These preliminary conclusions have some theoretical, practical, and
methodological implications, an analysis of which is presented in this chapter, in relation to
the literature and the possibility of generalizing the results, together with some future lines
of research opened by the limitations of this study.
Introduction
- 8 -
III. Research question and objectives
As was mentioned before, the research questions that guide this thesis aim to analyze the
relationship between the NIS and micro-heterogeneity and examine why some firms are
better prepared to deal with the unstable Argentinean environment. The underlying idea is
that the characteristics of such systems are not a temporary feature nor a previous stage to
development, but instead represent a structural situation. In the case of Argentina, abrupt
institutional and economic instability, deep changes in the rules of the game, and political
and social convulsion have been constant features over the last century. Given this history,
firms cannot be expected to behave as if there were a stable future. On the contrary, firms
are part of this system and they are cause and consequence of this particular environment.
Given the research questions, the objective of this thesis is to study innovative strategies at
the firm level in terms of how they persist or change over time. The research is focused on
the relationship between the NIS approach and the sequence of events brought about by
the change in the environment, firms’ reactions, and the impact of these reactions. The
objective is not to analyze how the selection process defines “winners and losers” but to
understand how, in a given environment, firms behave and respond differently in terms of
innovation and competitiveness.
The general objective is thus to study how innovative strategies change over time within an
unstable innovation system. The specific objectives (SO) are:
SO1: To make a contribution to the NIS approach in terms of how it deals with the
existence of micro-heterogeneity in unstable environments;
SO2: To identify and characterize different innovative strategies and their different impacts
on productivity, labor growth, and exports;
SO3: To make a contribution to the understanding of how innovation strategies work, in
order to propose key criteria for public policies capable of fostering successful innovative
behaviors.
IV. Methodological questions and hypothesis
Given the objectives, environmental instability will be contrasted with micro-regularities.
The methodology aims to identify and characterize continuities and discontinuities in firm
behavior and, especially, changes in this behavior over time. Chapter 4 deals with the
Introduction
- 9 -
operationalization of the objectives and concludes that, to empirically address the
objectives, the concept of “innovative strategy” will be used. This is an ex-post conceptual
tool that can be analyzed by looking at the regularity of the strategy (continuous or
sporadic investments in innovation), the complexity of the innovative structure (relative
innovation investment, capability, and resource levels), and the firm’s economic
performance (productivity, labor growth, and export levels). Chapter 5 deals specifically
with the translation of all these concepts into variables and presents the data.
The hypotheses aim to study the innovative strategies and arise from three different
methodological questions (Q), analyzed in depth in Chapter 4, but worth presenting here in
order to introduce the hypotheses (H) that will be tested with econometric tools. These
hypotheses are discussed and formalized in chapters 6 and 7, where they are tested on 800
Argentinean manufacturing firms for the period 1998–2006.
Q1: Is there a positive relationship between the continuity of the innovative strategy and
the complexity of the innovative structure? (Chapter 6)
H1.1: Continuous innovative firms have a more complex innovative structure.
H1.2: New innovative firms changed their innovative strategy in the direction of a more
complex structure over time.
H1.3: Sporadic innovative firms have a low-complexity innovative structure sustained over
time.
Q2: Is there is a positive relationship between the continuity of the firm’s strategy and its
past and present innovation results? (Chapter 7)
H2.1: The probability of innovating depends positively on having innovated in the past.
H2.2: The probability of innovating depends positively on the innovative strategy.
H2.3: The probability of innovating depends positively on having innovated in the
previous period when following a continuous innovative strategy.
Q3: Is there is a positive relationship between the continuity of the firm’s strategy, its
innovation results, and economic performance? (Chapter 7)
H3.1: Innovation results have a positive impact on firm performance (productivity, labor
growth and exports).
H3.2: Continuous innovative firms improve their performance improvements, regardless
of their innovation results.
Chapter 4: Towards a conceptual framework for innovative strategies
- 10 -
Section I - The National Innovation System
approach
Introduction to Section I
Since the publication of Schumpeter’s key works (1912, 1942), there has been a prolific
production of studies about innovation and its implications for firm performance and for
the development of the rest of society. According to these studies, innovation is an
endogenous variable of capitalism and is both cause and consequence of technological
progress.
The national innovation system (NIS) approach falls within these studies. In its most
simple definition, the NIS approach is a methodological tool for studying the process of
knowledge creation that leads to innovation (Lundvall, 1992; Nelson, 1992; Freeman, 1995;
Edquist, 2004). Within this approach, growth and development are the result of a systemic
process of competence building which enhances the capabilities of firms and individuals
and allows technological progress to take place. Of course, it is also an analytical tool for
explaining the obstacles that this process could face. In this sense, it allows key areas where
public policies could enhance or foster development to be identified.
Argentina falls within this latter use of the approach. During the last 20 years, several
studies have been carried out in order to explain the underperformance of this NIS. These
studies found a systematic lack of articulation between the different institutions and
organizations that make up the system, which has led to a poor innovative performance,
biased towards the acquisition of capital goods without endogenous efforts capable of
selecting, adapting, and improving them (Dutrenit and Katz, 2005; López, 2003; Lugones
and Suárez, 2006). This situation originates in the historical processes that shape the
Argentinean system and explains the existence of a dual productive structure made up of a
highly competitive agricultural sector that has remained a commodity exporter, on the one
hand, and a manufacturing sector that is highly dependent on public protection to survive
and that is oriented to domestic demand, on the other. What these studies do not show is
that there are firms which escape these standard explanations and show high-profile
Introduction to Section I
- 11 -
innovative behavior, linked to a high economic performance (Borello et al., 2005;
Chudnovsky et al., 2004; Katz, 1976; Kosacoff, 1998; Lugones et al., 2008). This micro-
heterogeneity (differences in innovative behaviors and productivity levels) seems to have
been an intrinsic element of Argentina’s NIS and begs certain questions about the
approach’s capacity for accounting for it and the relationship between the two findings: a
national system that disincentivizes innovation and the existence of successful innovative
firms.
Therefore, the objective of this section is to analyze how the NIS approach can answer my
research question. In the following chapters, I will discuss the structural causes that
account for the national innovative dynamic and the gap with the technological frontier but
also, and especially, how the NIS approach deals with the micro-heterogeneity and the
implications of this for long-run growth. It follows that the research questions for this
section have to do with the usefulness of the approach to explain both how historical
processes determined an unsuccessful innovation system but also how successful
innovative cases can be found even in this dysfunctional environment.
As I shall demonstrate, the NIS approach will allow the causes of the underperformance of
the Argentinean system in terms of growth and technological development to be identified.
Most of the innovative dynamics of Argentinean firms can be explained through the
historical processes of disarticulation between the productive structure, the knowledge
infrastructure, and public policies, within a context of recurrent cycles of recession
followed by uneven growth. However, we will also see that the approach falls short when
explaining the existence of firms that managed to successfully innovate and grow within
this environment.
This section is organized as follows. Chapter 1 analyzes the NIS approach, with the
objective of discussing its usefulness and limitations as a theoretical framework for
explaining innovation. In Chapter 2, the NIS approach is applied to the Argentinean case
with the objective of presenting why Argentina is an unstable environment and how the
NIS can shed light on the differential performance of firms. The rationale behind
examining the approach first and then the Argentinean case lies in the need to set out the
basic concepts through which the Argentinean case will be analyzed.
- 12 -
Chapter 1: From a “focusing device” to a
matter of public policy
Introduction
Since the publication of National System of Innovation: Towards a Theory of Innovation and
Interactive Learning (Lundvall, 1992), the concept of the NIS has gained wide acceptance
from both scholars and policy makers. In a brief historical review, this approach originates
in List’s (1841) concept of National Production System, and probably even two centuries
earlier, in the mercantilist ideas of Antonio Serra (Reinert and Reinert, 2003). In recent
decades, the systemic aspect of these ideas were taken up by Freeman in a report for the
OECD in 1982 (Freeman, 1982a) and were finally internationally disseminated through the
aforementioned work of Lundvall (1992), and through Nelson (1993) and Freeman (1987).
Theoretically speaking, the concept of the IS has two main elements: the idea of diverse
interrelated organizations (system) and the technological and organizational improvements
that they trigger (innovation). A third element is usually attached when it is translated into
practice: the geographical (national, regional, or local) or sectoral component. The systemic
component of the approach emerges as a critical response to the automatism of orthodox
economic theory, where the capitalist dynamics is understood as the result of a
simultaneous and linear addition of individual decisions in a rational world of perfect
information. The focus on innovation refers to the understanding of technological
development as a means for obtaining dynamic competitive advantages capable of fostering
development. Finally, narrowing the concept to the national, regional, local, or sectoral
level accounts for the importance of the time and place where organizations interact,
including the tangible and intangible institutions that regulate their actions. Originally, the
approach focused on the national level, which remains its most disseminated practical
application. However, the proliferation of other dimensions (specifically sub-national and
sectoral analysis) has led the main authors to generalize the approach by eliminating the
national focus from subsequent conceptual revisions. Since these revisions will be included
in this study, I will refer mostly to the NIS except in those cases where the contributions to
the approach refer explicitly to the IS, with no mention of the geographical dimension.
Therefore, the aim of this chapter is to analyze the NIS approach, its scope, and application
to countries with different levels of development. With regard to our research question,
Chapter 1: From a “focusing device” to a matter of public policy
- 13 -
this chapter will set out the theoretical framework from which the Argentinean situation
will be analyzed and will discuss the main contributions and limitations the approach has
when it comes to less developed countries.
This chapter includes the discussion of a set of elements that have contributed to the
consolidation of the NIS approach and the agreements and disagreements about its ability
to explain technological change. It does not intend to review all contributions that have
enriched the approach but to present a set of elements that account for its usefulness and
limitations as a methodological tool. Given the general objective of this thesis, special
attention will be paid to Latin America, given that, like Argentina within it, the region as a
whole has a history of instability.
The structure of the chapter is as follows. After this introduction, section 1.1 discusses the
origins of the IS approach, in order to present the basic concepts and premises. Section 1.2
reviews the approach itself, in an attempt to carry out a critical reading of its theoretical
contributions. The capacity of the approach to explain development is discussed in section
1.3, where the main agreements and disagreements for the case of Latin America are
presented and linked to the concept of instability. Finally, some preliminary conclusions are
provided.
1.1. Theoretical background: innovation economics
The Innovation Systems (IS) approach is part of the general framework called Innovation
Studies, which analyze, as the name implies, the phenomenon of innovation and
technological change. These studies seek to explain the determinants of the innovation
process and its impact on economic growth and development. Within this general
framework, the IS approach can be framed within the neo-Schumpeterian evolutionary
theory of innovation and technological change, although there are also certain theoretical
similarities with other economic approaches such as the analysis of industrial organization,
structuralism, and the new theories of international trade. Although this is essentially
heterodox economics, contributions from other social sciences and management, and the
explicit recognition of complementarity with historical, cultural, political, and
anthropological approaches, among others, make it one of the broader theories regarding
the explanation of development and the differential growth among nations. Of course, this
also makes the IS approach vaguer and more difficult to synthesize.
Chapter 1: From a “focusing device” to a matter of public policy
- 14 -
In stylized terms, the core of the IS approach is based on Schumpeter’s work about
innovation and its role within capitalist dynamics and Freeman’s and Nelson’s early
contributions on the same issue (Figure 1.1). From Freeman’s work, the approach
acknowledges the importance of demand and interactions with science and technology
(S&T) institutions and his criticism of Schumpeter’s exclusive attention on supply
(Freeman, 1974). From Nelson’s early work, the NIS approach takes up the concept of
routines and capabilities (Nelson and Winter, 1982). The original contribution of the
approach lies in its systemic view of the innovation process, where the knowledge supply,
producers, and users jointly determine its intensity and direction within a process which is
context- and time-specific, marked by historical determinants, formal and informal
institutions, and path dependence characteristics (the system) (Lundvall, 1992).
Figure 1.1: Main theoretical contributions to the NIS approach
1.1.1. The capitalist dynamic
The Schumpeterian root of the IS approach can be observed in the recognition that the
core of the capitalist dynamic is technological competition, through which firms search for
extraordinary profits. As defined by Schumpeter (1912), capitalism is an evolutionary
process where major changes occur through revolutions in which there is a process of
“industrial mutation” (qualitative change) that destroys products, processes, and companies
and creates new ones in a process of “creative destruction.”
The Schumpeterian creative destruction process begins with the development of a new
product or process by a firm and its successful introduction into the market (innovation).
Once the innovation has been made, it opens up imitation processes by other firms, which
in turn incrementally improve the product to achieve some differentiation. At the same
time, a new set of suppliers may emerge, leading to the emergence of entirely new
economic activities, which tend to compensate for the destruction of industries that now
Schumpeter
Innovation Studies
National Innovation System
Chapter 1: From a “focusing device” to a matter of public policy
- 15 -
become obsolete. The process begins again when a new radical innovation breaks into the
market.
Within this dynamic, Schumpeter refers to innovation as the key to capitalism, with either
the entrepreneur or the oligopolistic firm as the central player, depending on whether we
are referring to “Schumpeter Mark I” or “Schumpeter Mark II” (Freeman, 1982b). In the
first case, the entrepreneur, driven by the supply of science, decides to take risks to create a
new product or implement a new process—similar in many aspects to Keynesian animal
spirits (Keynes, 1936). In the second case, the key variable is the institutionalization of
research and development (R&D) within the firm, which determines the discovery of new
technologies.
This double reading of Schumpeter’s work (or the two points in time at which he wrote)
has led to the emergence of two concepts: the aforementioned “creative destruction” and
the concept of “creative accumulation”. Malerba et al. (1997) argue that in Theory of
Economic Development (Schumpeter, 1912) the key to the capitalist dynamics is the process of
destruction, which triggers innovation and leads to incumbents (firms competing in the
market) being displaced by entrants (new firms with innovative products). Instead, in
Capitalism, Socialism and Democracy (Schumpeter, 1942), the key process of capitalism is the
accumulation that arises from achieving a differential rent from an innovation. While in the
first case the central actor is the individual innovative entrepreneur, in the second it is the
large oligopolistic firm, which competes with other oligopolies trying to achieve a
differential rent.
Langlois (2002) and Yoguel and Barletta (2010) present a different interpretation, arguing
that Schumpeter never abandons the argument that the dynamics of capitalist development
is based on the process of creative destruction, although it may be independent from the
type of competition prevailing in the market. In some cases, the same oligopoly “destroys”
the market after the introduction of a new product; in other cases the entrant firms
introduce an innovation that challenges the market power of existing competitors.
Both dynamics (destruction and accumulation) are taken up by the IS approach, to which
later contributions by Schumpeterian and neo-Schumpeterian authors will be added: the
role of demand in triggering the innovation process (Schmookler, 1979), its iterative nature
and the importance of the firm’s interrelationships with its environment (Mowery and
Rosenberg, 1982), knowledge as a source of competitive advantage (Nelson and Winter,
Chapter 1: From a “focusing device” to a matter of public policy
- 16 -
1982), and the particular technological trajectories that determine the innovation process
(Dosi, 1982). In all these cases, innovation is at the center of the explanations and because
of this, a clear definition of this concept is required, which will also allow me to present the
rest of background to the IS approach and the main concept of this thesis.
It is worth mentioning that, there is a clear chronological order to some of the
contributions to the IS approach, if its origins are established as being 1992–1994,
following the publication of the main books on the subject (Lundvall, 1992; Nelson, 1993;
and Freeman, 1987; as discussed in the introduction to this chapter). In other cases,
however, the chronological order is not so clear. On the one hand, this is because the IS
approach has evolved since it first appeared at the beginning of the 1990s (Lundvall et al.’s
handbook on IS is from 2009). On the other hand, the aforementioned contributions to
the IS approach have also evolved—for example, Freeman and Soete (1997) is actually the
third edition of Freeman’s 1974 book—and will mostly be presented in their latest versions
or editions, in which concepts are clearer and better consolidated.
1.1.2. Defining innovation
A key Schumpeterian idea is that the dynamic of capitalism is endogenously determined.
This means that innovation is part of a process closely related to technological change and,
ultimately, to economic growth. In this context, innovation is defined as the development
of new products and processes (Schumpeter, 1912). Using this basic definition, the concept
has been applied at different levels. There is a macro level, usually associated with radical
innovations that transversally revolutionize existing practices (this is the case with the
internet, for example); a meso level, where innovations are defined as radical changes in
specific productive sectors (direct seeding and GM soy, for example); and a micro level,
where innovation means the implementation of new product, process, or organizational
techniques by a company (including, for example, imitation/improvement of a product
already marketed by the competition). In what follows, these different uses of the concept
are reviewed.
1.1.2.1. Innovation at the macro level
When he first used the term innovation, Schumpeter (1912) defined it as a new product or
process and its successful introduction into the market, as new ways of organizing business,
new sources of supply, or the exploitation of new markets. Once introduced into the
market, an innovation displaces old products and practices and remains a “novelty” until a
Chapter 1: From a “focusing device” to a matter of public policy
- 17 -
new innovation replaces it. In Schumpeter’s work, innovations are referred to as novelties
for the market that are introduced by a firm, on which the creative destruction process
subsequently operates (such that innovations that are displaced by others), giving way to
the classical mechanisms of competition that erode prices and profit margins.1 Thus,
although Schumpeter recognizes the importance of innovation and marginal changes at the
firm level, his approach is essentially macro and refers to radical innovations that break
with the status quo of markets, even creating new ones.
Three decades later, Schmookler (1979) took up Schumpeter’s ideas and differentiated
between process and product innovations in order to distinguish between changes in goods
and services from changes that directly impact how they are produced. This distinction
arose from Schmookler’s criticisms regarding the lack of participation of demand in
Schumpeter’s selection process and resulted in what was called later the demand-pull
approach (in contrast with the supply-push approach attributed to Schumpeter.)2
Also at a macro level, but with a clear impact on micro definitions, Freeman and Soete
(1997) distinguished between radical innovations and marginal or incremental innovations,
according to the degree of technological novelty involved. Similar to what was postulated
by Schumpeter (1912; 1942), the former involves a change in the way needs are met,
including the creation of new needs, and the latter a better way of meeting them.3. These
different types of innovation, however, are not only the result of entrepreneurial activities
or the competition dynamics of the oligopolistic firm—as Schumpeter might have
argued—but of the interaction between the firms’ behavior, the demand the firm faces (or
wants to face), and its relationships with the S&T infrastructure (Freeman, 1974). In this
sense, Freeman (1974) posits that other organizations (besides the firm) play a role in the
creation of innovations, and argues that investments in knowledge (beyond Schumpeter’s
biased focus on R&D) are a way of gaining extra profits based on innovations.4
1 From this explanation of the dynamics of the capitalist competition, Nelson and Winter (1982) derived their concept of selection process, which I will return to in Section II. 2 For Schumpeter, the entrepreneur identifies S&T inventions with the potential to be commercialized and then transforms them into innovation. Under this approach, the supply pushes the market dynamic when it introduces new products (supply-push) (Freeman, 1982; Kline and Rosenberg, 1989). 3 A similar differentiation was made by Mensch (1975), who distinguishes between radical and incremental innovations, like Freeman and Soete, but subdivides incremental innovations into incremental and pseudo-innovations, with the latter referring to those minor changes (e.g. packaging) that are not enough to exploit a niche but allows the firm to sustain its market share. 4 This led Freeman (1974) to identify ideal types of innovative strategies, depending on the combination of types of knowledge. Freeman’s work on this will be examined in Section II.
Chapter 1: From a “focusing device” to a matter of public policy
- 18 -
Within the IS approach, although the concept of innovation is not discussed, studies
associated with this macro level can be observed and the evolution of technology is
explained in connection with the evolution of the NIS. In this sense, the concept of
innovation is that defined in Schumpeter’s initial postulates, enriched by the work of
Schmookler, Freeman, and Soete. In other words, while Schumpeter assumes that there is a
constant flow of inventions (the knowledge supply) from which the firm will select the
profitable ones, for the IS approach the knowledge infrastructure will have a key role in
both the development of new products and processes and the formation of Schumpeterian
entrepreneurs (innovative firms).
1.1.2.2. Innovation at the meso level
At the meso level, the concept and understanding of innovation as technological change
can be identified in Dosi’s (1982) definition of a technological paradigm, Pavitt’s (1984)
taxonomy regarding the determinants of technological change, and Lall’s (1984, 1992)
contribution to the classification of industries according to their technological intensity.
Dosi' (1982) bases his idea of technological paradigms on the Kuhn’s notion of scientific
paradigms, but applies them to technological change. In this way, he argues that
technological progress follows paths that start with scientific or technological
breakthroughs, which once imposed determine the range of techniques and procedures
through which problems are identified and solved. With a similar meaning, Nelson (1994)
and Freeman (1995) explain that these trajectories are not only explained by technological
breakthroughs but by the co-evolution between firms, industrial structures, and institutions.
Technological innovation, in the sense of new technologies that have been successfully
introduced into the market, has to do with the new practices, materials, and techniques, but
also with the response from the demand, the organizations in charge of regulating
economic activity, the potential suppliers, and all other actors and rules (institutions in its
broad sense) that affect the economic dynamics of production, distribution, and
consumption.5
Also concerned about the role of S&T institutions, demand, supply, and producers in the
dynamics of technological change, Pavitt (1984) distinguishes between science-based
5 This particular point is where a blind spot in Nelson’s work about innovation at the firm level and the innovation system, at the national level, to the extent that while he acknowledges the impact of institutions on the innovation process, he refers to the selection process as an evolving interaction only between the firm, the supply and the demand (Nelson and Winter, 1982 vs. Nelson, 1994 or Nelson and Dahlman 1991). We will back to this in Section II.
Chapter 1: From a “focusing device” to a matter of public policy
- 19 -
sectors, scale-intensive sectors, supplier-dominated sectors, and specialized supplier sectors.
For this author, the technological dynamics of a country is determined by its sectoral
structure and how sectors fit together, accepting the possibility of the absence of one or
more sectors and thus the prominence of a technologically unbalanced productive system,
with strong barriers to development.6
Another contribution in this line is the taxonomy partially developed by Lall (1992) and
then taken up by the OECD (1997) according to the technological intensity of industries
(relative R&D investments). According to this classification, a productive structure is the
result of the particular composition of industries of high, medium, and low technological
intensity, which on aggregate determine the level of national technological development.
The greater the participation of high-tech sectors, the greater the technological intensity of
the system, and thus the level of income and industrial development. To some extent, this
classification seems to be a step backwards with regard to Freeman’s work on the role of
other sources of knowledge and Nelson and Winter’s (1982) work on the role of routines
and interactive learning. However, the simplicity of the classifications and the increase in
the availability of information on R&D for a large group of countries and sectors
contributed to its diffusion worldwide, as it is one of the most commonly used taxonomies
nowadays.
Coming back to the IS approach, Dosi’s technological paradigms and Lall’s and Pavitt’s
sectoral classifications were taken up by Malerba and other authors (Breschi and Malerba,
1997, Malerba and Orsenigo, 1997; Malerba, 2004; Malerba and Nelson, 2007; Cassiolato et
al., 2000; Cassiolato and Lastre, 2005, among others) to argue that the dynamics of
technological change differ between productive activities, which leads to the existence of
technological domains, with specific institutions, actors, and networks which lead to the
conformation of sectorial networks that transcend national characteristics. In this sense, the
Sectoral Innovation System approach argues that the dynamics of technological change can
only be apprehended on the basis of the identification of sectoral specificities (at the
national or regional level), which are associated with technological regimes determined by
specific conditions of appropriability, accumulation, the knowledge base, and opportunity,
of course at the sectoral level (Malerba, 2004).
6 Pavitt’s (1984) taxonomy is at the basis on many studies at the firm level, following the idea that the knowledge sources he identified applies not at the sectoral (meso) level but at the micro one (more on this on Section II).
Chapter 1: From a “focusing device” to a matter of public policy
- 20 -
1.1.2.3. Innovation at the micro level
As the concept of innovation was gaining ground in academia and politics, the need to
analyze it, characterize it, and give it direction led to the micro analysis of innovation being
consolidated and the more macro approaches becoming associated with the analysis of the
link between technological progress, growth, and development. Within the IS approach,
the concept of innovation at the micro level grew among empirical studies, which
associated it almost exclusively with the process that takes place within the firm as a result
of its relationship with the environment. In this regard, Dosi’s more recent work, Lall’s and
Fagerberg’s original contributions, and Nelson and Winter’s work are basic premises for
those who analyze innovation systems.
According to Fagerberg (2003), to innovate means “to combine different types of
knowledge, skills, abilities, and resources” so as to achieve improved technology which,
applied to products or processes, allows the firm to reach extraordinary profits. This
combination is not passive but involves making explicit investments aimed at improving or
creating skills and technological capabilities (Lall, 2001).
Another way of analyzing innovation at the micro level is that raised by Nelson and Winter
(1982). Faithful to Schumpeterian tradition, the authors explain the process of innovation
from the firm to its impact on the market structure and, ultimately, growth. Although they
do not propose a categorical definition of the term innovation, some elements of their
analysis allow an approximation to their interpretation of the phenomenon.
Part of Nelson and Winter’s (1982) central theoretical contributions is the definition of
innovation as changes in routines. For these authors, a firm is a set of routines, which are
more or less standard ways of “doing things” and constitute organizational memory.
Incremental improvements, solving problems in daily operations, and the development of a
new products lead to changes in routines, which allow the company to sustain or increase
its profit margins. In this context, the authors are thinking of phenomena that range from
incremental and adaptive innovations within the firm (although they are not referred to in
these terms) to the product innovations discussed above.
Later in the same chapter, and also as part of their definition of routines, the authors
explain that there are routinized ways of innovating, which constitute organizational
activities aimed at “solving problems.” As the authors emphasize, there is no certainty
about how and when the solution will be reached, but the way it is sought is a routine. Both
Chapter 1: From a “focusing device” to a matter of public policy
- 21 -
the “habit” of forming a task force to address a specific productive problem as well as the
percentage of its annual budget that a firm allocates to its R&D lab are routines to
innovate.7
In any case, the authors understand innovation as the result of the interactions between the
firm and the selection process, on the one hand, and the firm’s capabilities to deal with
both, the productive dynamics of the firm, and the environment, on the other. In this
respect, innovation at the firm level is part of a broader process of competition, subject to
path dependence, routines, and capabilities (Nelson and Winter, 1982), which is, to some
extent, closer to Lall’s and Fagerber’s understanding of the innovation process.
In this respect, Lundvall and Lam (2007) argue that innovation is in fact the application of
knowledge which can be created, appropriated, and combined by means of science-based
activities—in the sense of Schumpeter’s (1912, 1942) R&D investments—but also by doing
and using—as with Nelson and Winter’s (1982) routines—and by interacting with the rest
of the system where the firm operates (Lundvall, 2009; Jensen et al., 2007). As will be
examined in the following subsections, “the system” referred to by Lundvall et al. (1992)
transcends Nelson and Winter’s selection process and Freeman’s description of linkages
with the S&T infrastructure and demand. It has to do with the set of organizations and
institutions affecting the process of knowledge creation, appropriation, and diffusion.
Nowadays, the most widely accepted and applied concept of innovation within the analysis
of IS is that of the Oslo Manual (OECD, 2005). The worldwide distribution of this manual
as a methodological tool for measurement exercises in both developed and developing
countries and the need to construct internationally comparable statistical systems led to the
global acceptance of its definition of innovation (at least at the micro level).
According to the Oslo Manual, innovation is the “implementation of a product (or service),
process, marketing method, organizational practices, work organization or external
relations, new or significantly improved,” while innovation investments are those intended
to bring about an innovation (OECD, 2005). Given the above definition, to imitate is to
7 In some cases, the authors also talks about a “successful” innovation at the firm level, which is accepted in the market (demand) and generates a differential income. In this sense, the authors are referring to those product innovations that reach the market, accepting the possibility of innovations (novelties) within the firm that fail to be “successful.” Thus, at the micro level, innovation refers to the solution of a problem (which may be, for example, the need to replace a material) which succeeds when it is introduced into the market and therefore generates profits for the firm and investment opportunities for their competitors and other market players (imitation and diffusion).
Chapter 1: From a “focusing device” to a matter of public policy
- 22 -
innovate, since it implies something new for the firm. This recognition of innovation as a
mere improvement or something “new to the firm” has to do with the focus on the
microeconomic impact of technological change, in an attempt to explain the determinants
of the firm’s micro performance.8
This is the concept this thesis is based on: innovation at the firm level. As such, based on
the Oslo Manual definitions, this thesis will use the term “innovative firm” to refer to one
that invests in innovation, regardless of whether it achieves results; while a “firm that
achieved innovation results” denotes one that actually attained innovations (new products,
processes, organizational, or commercial practices). As shall be seen, this differentiation
between innovative firms and firms that achieve innovation results is key to understanding
the relationship between knowledge creation and application and the impact of this on the
firm’s economic performance.
The relationship between innovation and the firm’s dynamics is discussed depth in Section
II, however, it is worth briefly presenting here how the IS approach integrates these
concepts. In this respect, Nelson and Winter’s notions of routines and capabilities and
Freeman’s understanding of the innovation process will be taken up, together with
Lundvall’s broad view of sources of knowledge. However, instead of the selection process
that validates innovations (makes them profitable), the IS approach focuses on the systemic
nature of that process in which the limits between the firm and the rest of the system
regarding innovation processes are less clear. Defining “the system” and explaining why
“innovation” is a process are the objects of the following subsections.
1.2. The Innovation System approach
1.2.1. The four founding fathers: basic definitions
As was mentioned before, the innovation systems approach spread from the book edited
by Lundvall in 1992, which in turn was the result of the work of the Innovation,
Knowledge and Economic Dynamics group (IKE) at Aalborg University, Denmark. As
Lundvall himself explains, the IS approach is rooted in List’s concept of the Production
System (List, 1841) and first appeared in an unpublished report to the OECD written by
8 In practice, the definition proposed by the Oslo Manual has been taken up by those who analyze innovation at the firm level, without further discussions on the concept of innovation although with strong criticisms of the subjectivity present in the concepts of “novelty” and “significant improvements” and, in particular, its excessive attention to R&D investments to the detriment of other sources of knowledge creation, application, and appropriation.
Chapter 1: From a “focusing device” to a matter of public policy
- 23 -
Freeman in 1982 (Freeman, 1982a). Besides Freeman, two other authors directly involved
in the activities of the IKE group have also contributed to the definition of the main
concepts: Charles Edquist and Richard Nelson. Despite their joint work and many co-
publications, each has a particular position on the IS approach.
In the first edition of his book in 1992, Lundvall defines the NIS as “a social system
composed of elements and relationships that interact to produce, disseminate, and use new
knowledge, economically useful (...) located within the borders of a state nation or rooted
in that territory” (Lundvall, 1992). In this sense, Lundvall takes Schumpeter’s idea of
technological change as an engine of the capitalist system, but explains it through the
accumulation of knowledge that occurs within a system, in an evolutionary sense.
More recently, in the Handbook of Innovation Systems and Developing Countries, Lundvall et al.
(2009a) reinforce the complex and evolving nature of the NIS and define it as “an open,
evolving and complex system that encompasses relationships within and between
organizations, institutions and socio-economic structures which determine the rate and
direction of innovation and competence building emanating from processes of science-
based and experience-based learning” (Lundvall et al., 2009a). In this reviewed definition,
Lundvall’s is more concerned with the processes of competence building than with the
successful introduction of new products and processes. In this sense, he takes Fagerberg’s
(2003) ideas, assuming that the result (the new product in the micro sense of innovation) is
less important than the process of creating those innovations, because it is during the
innovation process that development-enabler capabilities are generated and accumulated.
In this sense, and emphasizing knowledge creation, Lundvall combines the importance
placed by Schmookler on demand and by Innovations Studies on knowledge, and argues
that societies build capabilities and grow because accumulation occurs throughout a
process that goes from the decision to innovate up to the erosion of earnings due to
market competition, and is shaped by other organizations apart from the firms themselves
and demand, and by other institutions in addition to the market. Moreover, some of these
will be related directly to the innovation process (e.g. S&T infrastructure) but others will be
transversal to the whole society (e.g. the education system).
Finally, the author considers that there is always an IS and that such systems should not be
compared, parameterized, or imitated without taking a large number of considerations into
account. His theoretical approach to the IS is in fact a methodological one, in that it is a
Chapter 1: From a “focusing device” to a matter of public policy
- 24 -
“focusing device” to address the reality that accounts for the determinants of innovation in
a specified place and at a specified time. There are no stages, phases, or recipes and, of
course, it is misleading to speak of “creating” an IS as it is a concept that arises ex-post
(Lundvall, 1992).
In 1991, as a prelude to Nelson’s 1993 work, Nelson and Dahlman define an NIS as “the
network of agents and set of policies and institutions that affect the introduction of
technology that is new to the economy” (Nelson and Dahlman, 1991). Like Lundvall, the
authors reinforce the systemic component of the approach and the importance of linkages
and cooperation. In this sense, Nelson and Dahlman not only take up the concept of
routines but also emphasizes the problem-solving logic that triggers innovative processes
within firms. Thus, the authors’ understanding of innovation resembles that of Freeman
(1974), to the extent that they accept the existence of different levels of complexity in new
products and processes and the incremental nature of innovation, but emphasize the
differential requirements of skills and connections with the environment.
An additional component of Nelson’s work (both with Winter in 1982 and with Dahlman
in 1991) is the explicit concern about the role of public policies that determine the rate of
creation and dissemination of technology and the differential development between
countries. In this sense, the determinants of catch-up and the technological gap stand out
over the innovation process in authors’ research agenda—Nelson’s in particular. In fact,
the subsequent theoretical and practical approaches of particular systems will be defined
based on these concerns.
Freeman’s (1987) analysis of ISs begins with his book on the Japanese case, Technology, Policy
and Economic Performance: lessons from Japan. In it, he argues that the NIS is “the network of
institutions in the public and private sector whose activities and interactions initiate,
import, and diffuse technologies.” A few years later, in The National Innovation System in
Historical Perspective (Freeman, 1995), the author takes up these issues and emphasizes the
historical processes that determine them. This work is perhaps the truest expression of the
Freeman’s approach and his recognition of the historical, cultural, social, political, and
economic determinants of the dynamics of a particular system and its position in the
ranking of development.
The author also argues that it is not merely the firm’s relationship with the environment
but the existence (or lack thereof) of a network of institutions that meet the technological,
Chapter 1: From a “focusing device” to a matter of public policy
- 25 -
organizational, and commercial needs of the productive structure in a coherent way
(Freeman, 2002). Freeman highlights the forces that at a given moment in a country’s
history and, given its relationship with the rest of the world, act on the system by means of
directing, limiting, or promoting technical progress and, with it, growth and development.
In this sense, his work entails a closer look at path dependence than Lundvall’s approach,
and a larger one in terms of the institutional determinants that Nelson highlights.
Edquist is another author who has contributed to the establishment of the approach. While
Lundvall’s goal is to develop a methodological tool to study NIS and Nelson’s is to identify
the determinants of differential development, Edquist’s concern is strongly associated with
pursuing the theoretical robustness of the approach. As he explains, the IS approach
cannot be considered a theory since it does not account for general and universal laws
(Edquist, 2004). He believes that the lack of homogeneity in key concepts (he even sustains
that Lundvall and Nelson refer to different issues with the same words) and the lack of
causal explanations and a clear conceptual framework make it a fuzzy approach, still in its
early stages of development. In his contribution to the Oxford Handbook of Innovation
(Fagerberg et al., 2004), Edquist proposes a set of standard features, applicable to all
systems, and strives to give a clear, accurate, and comprehensive definition: “a definition of
(national) innovation systems includes all important economic, social, political,
organizational, institutional and other factors that influence the development, diffusion and
use of innovations” (Edquist, 2004). He later adds: “Even a much more modest objective
of specifying the main functions of the IS, the activities and the components in them and
some important relations among these, would represent a considerable advance in the field
of innovation studies.” Thus, although the focus is on organizations and the relationships
between them in the pursuit of innovation, the functional definition of the IS places
innovation (results) at the center of the scene, as a pattern through which to measure the
success of the system. Consequently, his definition is a positive one based on an (arbitrary)
set of functions which, according to the author’s assumptions, are the mechanisms through
which an innovation is achieved. In this sense, Lundvall (2009) argued that the alternative
functions identified by other authors and the empirical evidence make Edquist’s list as
“limited and arbitrary” as Edquist’s criticisms of the lack of generality of the approach.
However, despite all these differences among the precursors of the approach, there are
some general consensuses. Firstly, there is a convergence space associated with the
systemic and complex nature of innovative dynamics and the agreement on the national
Chapter 1: From a “focusing device” to a matter of public policy
- 26 -
level as the one capable of explaining growth and development. Regarding the systemic
view, all authors agree that innovation is the result of the creation of skills within
organizations, the characteristics and possibilities of which are shaped by path dependence
learning processes, in a complex environment, which in turn explains the different levels of
development. At any point in time, there are forces driving (or blocking) the innovation
process, which are the result of institutional processes affected not only by economic
history but also by the social, political and cultural environment.
Regarding the national level, all authors state that it is within the nation-state that
innovation occurs and affects development, while other levels of analysis are
complementary approaches the sum of which determines the national aggregate. This
proposition is related to the emergence of the concepts of technological, local, and sectoral
innovation systems, which occurred simultaneously with the publication of the pioneering
works, sometimes as contributions to these books (Dalum, 1992; Breschi and Malerba,
1997; Malerba, 2004). In that sense, Edquist (2004) argues that the analyses of the sectoral
and regional levels are complementary to the IS approach and the selection of one over the
other depends on the analyst and the questions to be answered. Lundvall, meanwhile,
faithful to his initial methodological objective (to develop a tool for analysis), argues that
regardless of the chosen level of analysis, the concept of the IS is related to the recognition
of local embeddedness and tacit knowledge, which is generated and circulated within
geographical entities (Lundvall et al., 2009a; 2009b).
Finally, there is the concept of path dependence in the sense that “history matters” (David,
1985). Clearer in Freeman’s historical perspective and Nelson’s (1994) co-evolution of
sectors and institutions, the concept of path dependence has come to account for the
historical nature of innovation which, like any other social process, has to be
contextualized. At the same time, it accounts for the cumulativeness characteristics not
only of technology (mainly at the meso level) but also, and especially, of knowledge within
the firm and any other organization (Lundvall, 1992).
1.2.2. Building blocks for an IS: dimensions of analysis
From the previous section we can conclude that, in its most simple and broad definition,
an IS is the set of organizations, institutions, and relationships that account for the
innovative dynamics of a given environment at a given time, under the assumption that the
dynamic of the system will determine the direction and rhythm of the knowledge and
Chapter 1: From a “focusing device” to a matter of public policy
- 27 -
capabilities that lead to innovation. Given this definition, the application of the approach
demands the simultaneous study of components of the system and their shared history. In
this respect, four key concepts can be found: institutions and organizations, environment,
interactions, and competence building. (Figure 1.2)
There is a vast literature on each of these concepts, some of it directly related to the origins
of the IS approach. The number of contributions that have enriched the approach proves
that it is such a wide and complex theoretical background that there are as many key
concepts as research questions can be formulated. In this sense, it is worth mentioning that
the contributions that will be presented in the following sections are a selection of relevant
dimensions, but they are not a complete summary of contributions to the approach or a
literature review of all the discussions that have taken place within the framework. On the
contrary, they result from a selection of key elements with proven impact on the innovative
process which help to explain the relationship between the system and firms.
Figure 1.2: Key building blocks of the IS approach
1.2.2.1. Organizations and institutions
One set of contributions that shapes the concept of IS is related to the analysis of actors
and institutions. Throughout this chapter I have made explicit reference to the concept of
organizations, avoiding the use of the term “agents,” to the extent that this economic
concept usually refers to the smallest unit of decision-making regarding production,
Theoretical foundations
Schumpeter
Innovation Studies
National Innovation System
Interactions Environment Institutions &
organizations
Knowledge
infraestructure
Firms & other
organizations
Norms &
routines
National &
international
context
Competence
building
Main concepts
Chapter 1: From a “focusing device” to a matter of public policy
- 28 -
distribution, and consumption. This includes a number of neoclassical assumptions about
the determinants and direction of those decisions, which in turn implies the identity of
agents (the representative consumer/firm).
According to the IS approach, the concept of agent has a significantly different meaning.
Lundvall, Nelson, Edquist, and Freeman use the concepts of agents and organizations
interchangeably, and also use the term agent to refer to the minimum decision-making unit
(mainly firms and demand) but with less restrictive assumptions, even denying the fact that
there are two identical agents. For these authors and those who ascribe to their approach,
agents are individuals who act within a given environment with imperfect information and
bounded rationality, crossed by learning, competence building, and path dependence
processes. This concept is then distinguished from the concept of “institutions” (or “other
organizations”), which is usually used to denote all the other organizations within a system.
The concept of “institution” has also been widely discussed by economic studies, and a
wide range of meanings are associated with it. In the simplest neoclassical sense,
institutions are agreements between agents that regulate their behavior. The market is, of
course, the main institution (if not the only one). For most heterodox approaches, the term
has a similar meaning, but the existence of multiple institutions is recognized. In Lundvall’s
(1992) terms, institutions are “the rules of the game.”
According to the IS approach, a key contribution is Johnson (1992). The author assimilates
Nelson and Winter’s (1982) concept of routines to institutions, and states that they are
widespread habits shared by society in general or by a group in particular. Institutions lead
to behavioral patterns that reduce the rational calculations that an individual should make
when making decisions and as such, they lessen uncertainty. The relationship between
institutions and individuals is bidirectional: they arise as the routinization of individual
behaviors and once installed, they shape them. When “regularities in behavior” are
codified, regulated, and stored they are then formalized and converted into formal
institutions or organizations. However, and despite this clear differentiation between
organizations and institutions, Edquist (2004) draws attention to the fact that authors
connected to the IS approach will sometimes use “institutions” in Johnson’s terms
(collective habits, rules of the game), and sometimes as synonym for organizations.
Returning to Johnson’s analysis, like any product of human interaction, institutions are
permeated by political, economic, and social issues and influence future individual
Chapter 1: From a “focusing device” to a matter of public policy
- 29 -
behaviors. Thus, although all institutions are generally effective in the reduction of
uncertainty, they are not necessarily neutral (or “good”) in terms of the direction of the
decisions taken (Johnson, 1992). Surprisingly, even though both Johnson and Lundvall
draw attention to the need to look at power relations when the approach is applied to
developing countries (Johnson et al., 2003), this part of Johnson’s explanation (included in
Lundvall, 1992) does not play a key part in explaining the dynamics of any system, where
the distribution of power and interests among organizations and institutions within a
nation is neglected, except perhaps in Freeman’s (2002) analysis of the coherence of sub-
systems within national borders (such as the social, political, and scientific sub-systems).9
In terms of the role of institutions in the innovation process, one of Johnson’s central
contributions to the IS approach is the claim that institutions (formal and non-formal) may
be seen as drivers of technological change, as a counterargument to the “institutional
delay” hypothesis. According to this latter hypothesis, institutions lag behind technological
change, if not block it. To the extent that they are “ways of doing things,” institutions are
not prompted to change, and the development of new technologies may conflict with
existing agreements. Johnson states otherwise, arguing that institutions provide a regularity
that makes change possible. Although the author recognizes the impact of institutional
inertia and the extra economic and social costs produced by the perpetuation of obsolete
habits, he argues that institutions provide the framework within which problems and
solutions are identified. They allow individuals to find, improve, organize, and systematize
innovations within a technological trajectory (incremental innovations). At the same time,
as they provide behavioral habits that reduce the required amount of information, they free
up resources (the need to think) that can be allocated to alternative solutions (creativity),
leading to radical innovation (Johnson, 1992).
Within the scope of the IS there is no consensus on the sign of the impact of institutions
on the innovative dynamics of the system. Moreover, the impact of institutional inertia or
resistance to change is the subject of debate and argument related to the existence of
endogenous barriers to development, especially in the case of Latin American countries.10
9 Discussing the role of power relations and interests affecting the dynamics of the NIS is beyond the scope of this thesis, however, it is worth noticing that governmental elites and national bourgeoisies in Latin America have had played an important part in the establishment of the main institutions affecting the innovation process within the region (Galeano, 1998; Azpiazu and Notcheff, 1995). 10 See for instance Rivera Ríos’s (2009) work on the impact of institutions and the contributions of Bisang (1993) about the divergent paths between Argentinean firms and institutions (we will back to this in the next chapter).
Chapter 1: From a “focusing device” to a matter of public policy
- 30 -
In fact, Lundvall himself, a decade later, pointed out institutional problems as systemic
failures that explain the different levels of maturity of the NISs in different parts of the
world (Lundvall et al., 2009b).
1.2.2.2. The firm
Given the centrality of innovation and the fact that it mostly takes place within the firm, an
important number of contributions to the IS approach have to do with the explanation of
the micro dynamic of the innovation process. Moreover, firms are usually placed at the
center of the NIS. In this sense, the impact of the Schumpeterian tradition and a long
history of theoretical and empirical observation from the initial authors cannot be
ignored—indeed, Freeman’s and Nelson’s work regarding firm behavior were published
long before the first books about the NIS.
Since this subject will be discussed in depth in Section II, only two issues are worth
mentioning here, which both summarize how the IS approach defines the firm: the
conception of the innovation process as an interactive phenomenon and the importance of
decisions at the firm level. The former refers to the recognition of innovation as a process
that occurs within the firm but in relation to the environment and the competitive process;
the latter to the abandonment of the assumption of hyper-rational maximization in a world
of perfect information and the existence of diversity (as opposite to representative firms).
In relation to innovation as a systemic process, the IS approach recognizes the iterative
nature of innovation, to some extent based on Nelson and Winter’s (1982) ideas about
interactive learning but also on Kline and Rosenberg’s (1989) and their criticisms of the
Linear Model of Innovation.11 The authors explain the interaction between the supply-push
and demand-pull forces and argue that innovation is an iterative process between the
development of knowledge, the needs of the market, and the real possibilities of producing
goods.
This position is the predominant one among those who ascribe to the IS approach, who
also recognize that the innovation process depends on the firm’s ability to generate new
knowledge, which in turn results from the ability to absorb and recombine external
11 The Linear Model of Innovation defines the innovation process as a series of four sequential steps: the first three are the different activities that make up what is now called R&D (basic research, applied research, and experimental development) and the fourth is the innovation itself and includes all activities leading to the introduction of new products into the market and/or the implementation of new processes (Rossegger, 1987).
Chapter 1: From a “focusing device” to a matter of public policy
- 31 -
knowledge. In other words, innovation is an interactive process where knowledge from
different sources (R&D, interactions, experience, training, etc.) is created, combined, and
applied. The impact of these efforts in learning and improving will be mediated by the
firm’s ability to absorb and increase knowledge generated elsewhere (in the sense of Cohen
and Levinthal, 1990) but also by the firm’s endogenous capabilities (Nelson, 1991a, b).
In relation to business decisions, the IS approach accepts the existence of distinct
behaviors between agents, which arise from different decisions.12 Within this general
framework, the innovation process is defined as the result of entrepreneurs’ explicit
decisions, aimed at achieving technological or organizational improvements. Furthermore,
the existence of different types of innovation and different ways of achieving them leads to
the existence of different behaviors associated with different strategic trajectories (Lall,
2001; Fagerberg, 2003).
Since technological progress within the firm is a process with path dependence
characteristics that also depends on the firm’s ability to identify, develop, and adapt
technology, decisions taken today have impact on future decisions, and thus on the path of
technological change. The result of these decisions, in turn, determines the accumulation
process within the firm and its possibilities for sustaining differential rent (Nelson and
Winter, 1982).
The direct consequence of the existence of multiple behaviors is what Lundvall refers as
the “variety of the system” (Lundvall, 1998) and it explains the fact that no two systems are
identical in either morphology or competitive dynamics. Similar conclusions arise from the
routine-based explanations of Nelson, where specific combinations of skills within the firm
will lead to specific combinations of routines and, therefore, to differential behaviors and
results.
Although the discussion regarding the firm’s innovative behavior will be widely discussed
in the next section, it is worth advancing some limitations of the IS approach in explaining
such behavior. Firstly, the emphasis of the approach in explaining technological
development has led to a bias in the empirical analysis towards explaining firms’ innovative
dynamic as if, when deciding how to compete, firms most frequently opted for innovation
12 In a Schumpeterian sense, the creative destruction process presents different strategic possibilities, depending on when the firm enters the process, and will require distinct investments and decisions whether the firm is seeking to be the first, the second best, to take advantage of a growing demand within an emerging market, or to maximize profits from income at the stage of falling costs.
Chapter 1: From a “focusing device” to a matter of public policy
- 32 -
as the means to do so. In this line, there is little explanation of how a firm is able to survive
without investing in technological or organizational improvements.
The second limitation is derived from the latter. The Schumpeterian creative destruction
process is accepted as part of the approach and firms with lower productivity levels are
supposed to disappear from the market. The evidence, however, suggests that in some
markets this selection process is weak, which means that markets are made up of firms with
different productivity levels. In this respect, Salter (1960) argues that productivity depends
on the firm’s ability to reach scale economies, to appropriate new knowledge (sectoral
technological change), and to substitute factors. However, this ability will be subject to the
actual possibilities of it being realized, which depend on the institutions—in the sense of
Johnson (1992)—that affect the interactions between the firm and the environment: labor
regulations affect labor-capital substitution, regulations of imported capital goods affect the
appropriation of technological change, regulations of prices and taxes generally speaking
affect the cost function, etc. All these elements will impact the cost function of the firm,
which in turn will take decisions regarding the acceptable profit rate when facing the
market price (Nelson and Winter, 1982). As a result, the “market” is in fact the place where
organizations with different levels of productivity converge and is not a temporary
phenomenon but a permanent structure (Bottazzi et al., 2010).
This micro-heterogeneity impacts decisions about the innovation process within the firm.
As some less productive firms are also part of the market, the higher profitability levels of
the most productive enterprises are not eroded, and so new incentives to innovation
cannot appear. As a consequence, the market will be made up of a heterogeneous set of
agents, not only in terms of Lundvall’s variety or Nelson’s skills and behaviors but also in
terms of the productivity level.
This brings up a third limitation of the approach when applied as an ex-post
methodological tool: the existence of micro-heterogeneity associated with different levels
of productivity determined by different competition mechanisms. As was mentioned
before, the existence of a diversity of agents is at the core of the approach. However, when
this diversity of situations is associated with competition-damaging mechanisms, such as
unregistered work, tax evasion, enclave economies, or simply the leveraging of promotional
benefits without adding value (assembly), then it is necessary to analyze heterogeneity at the
micro level and its impact on the economy as a whole. Of course, the study of virtuous
situations is also required. According to the IS approach, the situations at each extreme can
Chapter 1: From a “focusing device” to a matter of public policy
- 33 -
be explained on the basis of individual trajectories and specific interactions with other
environmental organizations and incentives. However, accounting for both situations at the
same time is not possible. In that sense, it seems that the emphasis on the history of the
system has worked against its capability to explain differential reactions.
1.2.2.3 System and environment
Together with the concept of institutions, the “environment” is another fuzzy building
block of the IS approach. The use of the term “environment” implies that an
organization/system is at the center of the analysis while all other organizations are
considered only if they affect its performance. In this context, the environment is
understood as the set of public and private organizations that impact the innovative
behavior of firms and systems. The focus is not on society generally speaking but on those
institutions that affect the innovative dynamics in a particular place and at a particular time
(Lundvall, 1992; Nelson, 1993; Edquist, 2004). This implies that the institutions that should
be included in an analysis at a given moment may be different from those included at other
moments or locations. There are, however, some institutions which are always important.
In Lundvall’s (2009) work, the IS is considered in both a narrow and a broad sense. In the
narrow sense, the IS is made up of the knowledge infrastructure and the firm, which the
author refers to as the core of the system. The knowledge infrastructure (or knowledge
supply) includes universities and S&T institutions, whether they are public or private.
These institutions constitute one of the axes of the approach: universities in their role of
“producers” of skilled human resources and S&T institutions as knowledge creators.
In the broad sense, Lundvall refers to the system as the “wider setting” that surrounds the
core of the system. This extended concept includes all those institutions—formal and non-
formalized—that affect not only the innovative dynamic but also the process of
competence building: the labor market, the macroeconomic trajectory, the financial system,
the government, public and private demand, social dynamics, etc. Therefore, in Lundvall’s
view, some institutions for part of the narrow system while others are the wider setting.
Surrounding the system is the environment. To a certain extent, there is some overlapping
between Lundvall’s broad definition of the system and the concept of the national
environment, while the borders are clearer when the international context is taken into
account.
Chapter 1: From a “focusing device” to a matter of public policy
- 34 -
The national environment of an NIS is the set of institutions that, although not directly
linked to the innovation process, affects the decisions and actions of the organizations
within the system (e.g. macroeconomic policies or income distribution). From a broader
view, the environment is also the international context and the position the system has in
world capitalism, which also impacts the opportunities and constraints that the system
faces in terms of innovation.
Like Lundvall, Nelson describes innovation at the firm level as depending on technological
capabilities and, in this sense, the system is closely related to the S&T infrastructure
(narrow system, defined as a set of institutions). However, he also strongly focuses on the
macroeconomic environment—and the economic policy that determines it—as a key
aspect to explaining innovation. This author states that economic performance is explained
by the dynamic of the IS (firms and the knowledge infrastructure), the social absorption
capabilities, and the macroeconomic environment. The combination of these three
dimensions explains technological progress (Nelson and Dahlman, 1991). In this way,
Nelson—like Lundvall—understands the NIS in narrow terms although he specifically
adds that the macroeconomic environment is a key aspect of the context that surrounds
the firm and explains a country’s trajectory.
In this case, the distinction between the environment and the system is much clearer.
Nelson’s macroeconomic environment implies the existence of different dimensions of the
same environment associated with the main institutions that affect the innovation process
(the social, political, and external environments, etc). In the terms of this approximation,
firms would be part of the productive environment and the innovation system is the set of
firms and knowledge organizations (knowledge infrastructure in Lundvall’s terms) that
determine the rhythm and direction of innovation and technological change.
Freeman, in turn, sustains that the environment is related to the evolution of the set of
incentives, which tend to vary according to the economic cycle and the international
context (Freeman, 1995). Throughout Freeman’s work, there are clear references to the
context in which firms operate in, a very broad sense. He includes not only national
determinants, particularly public policies and the social and political context, but also the
country’s global position in terms of trade and power relationships. Similarly to Nelson,
Freeman argues that the innovation system is the set of firms and S&T institutions which
co-exist with other sub-systems of the national context (social dynamics, the political
arena—Nelson’s environments) and argues that the level of coherence between the IS and
Chapter 1: From a “focusing device” to a matter of public policy
- 35 -
the rest of the sub-systems affecting the economic dynamics determines not only the
rhythm and direction of technological change but also its success in terms of economic
growth and international competitiveness (Freeman, 2002).
Besides the differences between the authors, each approximation has strengths and
weaknesses, which become evident when the approach is applied to a concrete system. In
terms of the strengths, the emphasis on the active role of the S&T and educational
infrastructures is one of the key contributions to the technological dynamic. In this sense,
the interaction between the S&T infrastructure and firms and how the national context
impacts on them extends beyond a simple linear understanding of technology as a supply-
push or demand-push. On the contrary, learning is interactive (bidirectional flows) and the
particular set of incentives each organization faces and their level of capabilities will
determine the extent and complexity of the flow of knowledge and even the lack of a
relationship.
The main weakness seems to be the interpretation of the environment and the restricted
view of the context surrounding the firm and how far it impacts the innovative process. If
the innovation process depends on the firm’s decisions about how to face competition, and
the environment is the set of organizations and institutions affecting those decisions (and
the competition itself), then there could be multiple institutions affecting innovation. A
similar reasoning is observed by Johnson et al. (2003) who emphasize the importance of
the wider setting when the approach is applied to developing countries. I would claim,
however, that this is not just a feature of developing countries but a matter of how the
environment that affects capitalist competition shapes the innovation process regarding the
level of development of the region or nation.
In this sense, Ludnvall’s distinction between the narrow and broad system turns out to be
pointless since the challenge for analysts will be identifying the main forces that impact the
innovation process and then understanding the relationship between the firm and its
environment. For instance, a commercial policy that impacts the price of imports will
generate incentives to substitute them and therefore, interact with clients, suppliers, and the
knowledge infrastructure during the quest for the required innovations. As such, although
knowledge creation will take place in the “narrow system”, what triggered that process
were not only those institutions but also the “wider setting.”
Chapter 1: From a “focusing device” to a matter of public policy
- 36 -
Conversely, if the environment is the set of institutions and organizations affecting the
innovation process within the firm, then this concept is useful for accounting for the
achievement of new product, processes, or organizational practices within the firm but falls
short when explaining how these novelties become innovations (how they are introduced
into the market). Elements of the environment such as income distribution, prices
regulations, and even tastes and preferences will impact not only the commercialization of
innovations but also—and especially—earlier decisions to seek this innovation at all.
By way of summary for this subsection, in this thesis the national environment will be
understood as the set of institutions and organizations affecting the process of production,
consumption, and distribution, while the system is the set of institutions and organizations
affecting the innovation process (and is therefore a sub-system of the environment). As
shall be discussed in Chapter 3, this distinction only has theoretical value when the
incentives that emanate from the environment affect the firm’s innovation process so
deeply that differentiating between Lundvall’s broad definition of the system and the rest
of the firm’s environment is not possible. This means that a broad interpretation of the
system is required when it comes to explaining the different innovative dynamics and
results at the firm level.
1.2.2.4. Interactions within the IS: cooperation and interdependence
According to the IS framework, firms are interconnected and this interconnection works as
an information channel, in addition to the traditional trade-based ones, from which the
firm absorbs and generates knowledge. There are two types of interaction present in all the
theoretical work of the quoted authors: those that occur as a result of the firm’s operations
(the clearest examples are commercial interactions) and those made in order to reach an
innovation. This distinction is seldom made explicit but both are used to explain the
dynamic of any system.
When the goal of the interaction is the quest for technological or organizational
improvements, interactions are known as linkages for innovation. Among these, the
distinction is usually made between informal linkages and formal interactions, or
cooperation. The former occur without the mediation of explicit agreements or contracts
involving commitments of the parties. The latter are explicit and formalized cooperative
agreements, which often involve monetary commitments. In all cases, interaction acts as an
information channel and impacts the firm’s innovative dynamics.
Chapter 1: From a “focusing device” to a matter of public policy
- 37 -
Besides of the evolutionary imprint associated with the interaction among organizations,
the focus of IS analysis has drawn on several approaches that sustain that firms operate
within a system, so innovation in isolation is assumed to be impossible (Lundvall, 1992).
However, this is not new or exclusive to the IS approach. Studies on industrial
organization, such as Williamson’s (1981) work on transaction costs or Meade’s (1952) on
externalities, recognize the existence of less flexible relationships determined by extra-price
issues.
The contribution of the IS approach lies in the aforementioned systemic nature of the
interaction. In this sense, three issues will help to understand the focus of the approach.
The first is that the interaction is cause and consequence of organizations’ decisions, in
which the firm’s actual place within the network depends on its ability to interact beyond
the market mechanism. This network constitutes a key variable to understanding the set of
incentives the firm faces and, ultimately, its innovative dynamics. This is especially
emphasized by Lundvall, referring to the modes of innovation (Jensen et al., 2007); by
Nelson (1991a, 1994), when he argues that innovation is the result of firms’ routines and
the process of capitalist competition, which includes not only competitors but also the rest
of the institutions that regulate it; and by Freeman (2002), when he discusses the
importance of the coherence of the articulation and interactions between firms and
institutions.
The second issue is related to the fact that interaction enhances performance when the
organizations involved have reached a minimum level of competencies. The threshold of
capabilities is given by the organization’s ability to understand and absorb the knowledge
that circulates in the environment, which requires the understanding of the language, the
formal and informal mechanisms of transmission, and the parties’ motivations to interact.
The characteristics of the knowledge that the organization wants to absorb will determine
the level of the threshold (Cohen and Levinthal, 1990; Narula, 2003). In this respect, the IS
approach emphasizes the role of capabilities and competence building as key determinants
of the knowledge creation and application that leads to innovation (Lundvall, 2009).
The third issue associated with the dynamics of linkages within the IS approach is that it is
inevitable. In this sense, the dynamics of the capitalist system take place in environments
governed by institutions that define rights and obligations, and firms have to interact with
them in order to reach the market. This takes us back to the impact of the environment
and the multiple institutions affecting the innovation process.
Chapter 1: From a “focusing device” to a matter of public policy
- 38 -
Finally, an implicit assumption in the approach accepts that the more interconnected with
the IS the firm is—the denser its network of connections—the greater its chance of
absorbing knowledge from the environment and, therefore, achieving innovations and
improving performance. Other heterodox approaches, especially Complexity Theory and
Social Networks Analysis, discuss and also refute this assumption since evidence would
suggest that a key position within the network (the system, in the terms of the IS approach)
is not necessarily associated with the density of linkages.
1.2.2.5. Knowledge and competence building
According to the IS approach, and specifically, to Lundvall, learning implies adaptation and
competence building (Lundvall et al., 2009a). The former refers to the agents’ capability to
respond and adjust their behaviors to external impulses; the latter is related to the firm’s
learning capabilities and is a key aspect when it comes to know-what, how, who, and why.
In the early views of Edquist (2004), competence building was a narrow concept referred to
exclusively as a means of innovation. For this author, learning was one of the main
functions of the IS. The concept included education and training aimed at creating human
capital and reproducing skills to be used in innovation and R&D.
Freeman (1974) recognized the importance of knowledge long before the IS approach and
sustained that the key feature of the new economy is the acceleration of technological
progress, which is the result of the institutionalization of knowledge creation. To compete,
firms must be able to generate and transform knowledge, and this only happens if
resources are allocated to increasing skills and human capital.
Nelson’s view is similar: he places skills at the center of routines and the firm’s competitive
dynamic, explaining that the firm’s level of competences (the system) is more than the
aggregate of individual skills (organizations) (Nelson and Winter, 1982). He argues that the
technological capability is key to understanding the innovation process and the way
societies create and transform knowledge. Moreover, he explains that technology involves a
body of knowledge which in the modern era cannot be found in one single person but
rather in many. As a consequence, firms can innovate (apply new knowledge) by means of
interaction.
According to the IS approach, this implies that knowledge generation and accumulation
depends on the performance of individual efforts but also on the interconnections between
Chapter 1: From a “focusing device” to a matter of public policy
- 39 -
them. When Lundvall refers to the accumulation of competencies, Nelson to improved
routines, and Edquist to specific skills for developing countries, the authors are including
not only the actual capabilities of individuals and systems but also the set of collective
competencies that allow knowledge to be transformed into innovations.
1.3. The IS approach and unstable environments
1.3.1. Development as a relative concept
Since the first publications on the IS approach, its focus has been heavily influenced by
concerns about growth and development. These concerns have led to a fluid interaction
between researchers from developed and developing countries as well as to a large number
of cross-country empirical analysis. Discussing the causes and consequences of
development is beyond the scope of this thesis. However, it is worth providing a brief
discussion of the underlying idea of development within the IS approach and its
relationship with the concept of instability.
Theoretically speaking, two key contributions to the IS approach can be found in the
literature: Amartya Sen and Gunnar Myrdal (see for instance Lundvall, 1992; Johnson et al.,
2003). The former refers to development as freedom, the freedom of individuals to choose
and decide about their life (how to live, where to work, what to study, etc.), associated with
equality of opportunities (Sen, 1999). The latter refers to development as “the movement
upward of the entire social system” (economic and non-economic dimensions of life in
society) which is the result of a cumulative causation between economic growth and
equality (Myrdral, 1974). According to these definitions, the process of competence
building emphasized by the approach constitutes a means for improving individual
capabilities and national income by means of innovation. Therefore, it contributes to
development.
From an empirical perspective, development seems to be about the density of the network
of organizations that lead to innovation (and economic growth) and the level of national
income (in terms of the population) (see for instance Lundvall et al., 2009b and Nelson and
Dahlman, 1991). The former is related to the interaction between organizations and the
actual existence of organizations that allow the competence building process to take place
(firms, universities, public policy organizations, S&T facilities, etc.), which, according to
what was stated in the previous paragraph, is the source of growth with equity. The latter is
perhaps the most common measure of development (per capita GDP), not just because it
Chapter 1: From a “focusing device” to a matter of public policy
- 40 -
constitutes an average income level but because of its strong correlation with other key
variables associated with social welfare (incomes, literacy, health) and economic growth
(competitiveness, productivity, employment).
Based on any of these variables, one can establish a ranking of countries and notice the
existence of countries with higher levels than others. Those in the higher positions are
usually called developed countries, while those at the bottom are called developing ones.
Wherever one traces the distinction between the top and the bottom of the ranking, there
is no doubt that some countries are more developed than others, although correlation
between GDP per capita and the different dimensions of welfare and growth will not be
the same for each country (Cuba, for instance, has low maternal and child mortality levels
and low GDP per capita levels, while Brazil has relatively high levels of both GDP and
poverty), In these terms, Latin American countries can be said to be less developed than
European or North American ones, and African countries as less developed than all of
these (except perhaps South Africa). Along this gradient of less and more developed
countries, there is a similar gradient associated with the process of competence building
and innovation, which is in turn correlated with economic growth and social welfare.
Graphs 1.1 and 1.2 schematize this relationship in a very simplified but graphic way.
In the case of the relationship between GDP and GINI, countries from the European
Union (EU) show high levels of income and equality while countries from Latin America
present the opposite situation. The correlation is even higher in the case of the relationship
between GDP per capita and R&D investments, where once again, countries from the EU
show high levels for both indicators while Latin American countries present lower levels of
income and investment.
To conclude, this is precisely the meaning I will give to the category of developing
countries in this thesis. Firstly, it is a relative measure. Secondly, it a classification that
accounts for a set of elements associated with economic growth with equity. Thirdly,
innovation and competence building are means to this ultimate goal. Fourthly, Latin
American countries are developing ones in comparison with those from Europe and North
America. Fourthly, although African countries could be seen as less developed that Latin
American ones, it makes no sense to look at Africa in order to analyze Latin America’s
development possibilities.
Chapter 1: From a “focusing device” to a matter of public policy
- 41 -
Graph 1.1: Gross domestic product (GDP) and income distribution (GINI)- 200813
GDP p/c: per capita gross domestic product in thousands of PPP USD. GINI index. Source: WB (2012), UIS-UNESCO (2012) and Eurostotat (2012).
Graph 1.2: Gross domestic product (GDP) and expenditure on R&D (GERD) - 2008
*GDP p/c: per capita gross domestic product in thousands of PPP USD. GERD: gross expenditure on R&D to GDP (%). See footnote 14 for the countries included in each region. Countries and regions in footnote 13. Source: WB (2012), UIS-UNESCO (2012).
13 Year 2008 except from * and ** which correspond to year 2009 and 2010 respectively. Asia: Afghanistan, Azerbaijan, Bangladesh**, China, Egypt, Fiji*, Georgia, India, Indonesia, Iran (Islamic Republic of), Iraq, Israel, Japan, Jordan, Kazakhstan, Kyrgyzstan, Lao People's Democratic Republic, Malaysia*, Mongolia, Nepal**, Pakistan, Philippines*, Republic of Korea, Saudi Arabia, Singapore, Sri Lanka, Thailand, Tajikistan*, Viet Nam. Africa: Cambodia, Central African Republic, Burkina Faso, Côte d’Ivoire, Madagascar**, Mali**, Mauritania, Mozambique, Niger, Nigeria**, Uganda*, South Africa*. Europe (non-EU): Slovakia, Belarus, Serbia, Montenegro, Armenia, Albania, Republic of Moldova, Bulgaria, Romania, Russian Federation, Bosnia and Herzegovina, Slovenia. European Union: Austria, Belgium, Croatia, Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Italy, Latvia, Luxembourg, Malta, Netherlands, Norway, Poland, Portugal, Spain, Sweden, Switzerland, Turkey, Ukraine, United Kingdom. Latin America: Argentina, Bolivia (Plurinational State of), Brazil, Chile, Colombia, Costa Rica, Dominican Republic, Ecuador, El Salvador, Guatemala, Guyana, Haiti, Honduras, Mexico, Nicaragua, Panama, Paraguay, Peru, Trinidad and Tobago, Uruguay, Venezuela (Bolivarian Republic of).
20
25
30
35
40
45
50
55
60
65
0 10 20 30 40 50 60
GDP p/c
GINI
Asia Africa Europe (non-EU) European Union Latin America
0
0.5
1
1.5
2
2.5
3
3.5
4
0 10 20 30 40 50 60
GDP p/c
GERD
Asia Africa Europe (non-EU) European Union Latin America
Argentina
Argentina
Chapter 1: From a “focusing device” to a matter of public policy
- 42 -
1.3.1.1. Development and instability
Like the concept of development, instability is also a relative term. Within Europe, Italy is
probably more unstable than Germany, and within Latin America, Argentina is more
unstable than Uruguay. On average, Latin American countries are more unstable than
European ones. However, from the point of view of the IS approach, the international
environment of Argentina is not the same as the international environment of Italy and the
network of organizations and institutions capable of fostering the latter’s recovery (e.g. the
European Union) is much denser and more integrated than the environment Argentina has
faced during its different cycles (e.g. the MERCOSUR or IMF grants). The different
networks each country is inserted in as well as the national system explain part of the
different trajectories schematized in Graphs 1.3 and 1.4 for a selected group of countries.
In terms of the evolution of the GDP, while South American countries present an erratic
tendency with oscillations between 0.7 and 1.3 with regard to the 1980 value, European
countries show an increasing tendency that is sustained over time (Figure 1.5). The
instability is clearer if different macroeconomic indicators are compared. In the comparison
between Argentina and Denmark (Figure 1.6), the exchange rate presents a completely
different scale, explained mainly by Argentina’s hyperinflation at the end of the 1980s.
The most common use of the concept of instability within innovation studies refers to
economic instability, associated with economic cycles, but also with the dynamic of the
interaction between aggregate demand and supply and fiscal accounts (this is precisely what
was observed in Graphs 1.3 and 1.4). Social, political, and institutional instability are also
analyzed but, most of the times, only in relation to their interaction with the economic
cycle—as causes (uncertainty) and consequences (recession)—(Nelson and Dahlman,
1991). From the point of view of the economic cycle, one can think again in terms of a
gradient between long and short cycles and between deep and soft oscillations. Thus,
countries with economies that have long cycles/soft oscillations are more stable than
countries that have short/deep ones.
Of course there is a link between economic and institutional (social, political, legal, etc.)
instability in that this increases the level of uncertainty, deteriorates welfare levels and
impacts the freedom of individuals (Ocampo, 2012). However, while there is no linear
association between the macroeconomic situation and innovation, the impact of
institutional instability on innovation and competence building is most probably negative.
Chapter 1: From a “focusing device” to a matter of public policy
- 43 -
Graph 1.3: Gross domestic product (1980=1)—Selected countries (1980–2006)
Source: WB (2012).
Graph 1.4: Argentina and Denmark (1980-2006)—Selected indicators (1983=1)
GDP p/c: per capita gross domestic product in constant 2005 international USD. Unemployment: % to labor force. Inflation: consumer price index, annual %. Official exchange rate: local currency unit per USD, annual average (secondary Y axis). Source: WB (2012).
With regard to the relationship between macroeconomic instability (short economic cycles)
and innovation, Antonelli (2008) modeled the reactions of firms to macro incentives in a
U-shaped paraboloid pattern, which links benefits and time. On the left are firms with
lower productivity values, in the center those with average levels, and to the right firms
with a higher than average level of productivity. Applied to the economic cycle, this
relationship implies that during the downturn, incentives to innovate are given by the need
for survival; in the stable stage, incentives are given by the need to differentiate the firm,
escaping from neoclassical perfect competition; and during the expansion phase, the
availability of funds allows the firm to take risks, face more radical innovations, and afford
the costs of experimentation that companies in less favorable positions cannot.
Consequently, any point in the economic cycle has the potential to create incentives to
innovate.
A more traditional interpretation of the impact of the economic cycle argues that during
recession, unemployment levels help to reduce controls on registered labor, which added to
Chapter 1: From a “focusing device” to a matter of public policy
- 44 -
the economic suffocation, creates incentives to base survival on non-registered schemes or
precarious economy, which adversely affects the innovative dynamics. The stable phase,
meanwhile, is a “desirable” state in which the firm can maintain its profitability without
taking on technological or financial risks, which also have negative impacts on incentives to
innovate. Finally, the expansion phase is a time of growth pulled by increasing demand,
when pressure on prices decrease. In this case, there would be no incentive to innovate
since profits can be obtained by structural effects.
Consequently, predicting the firm’s response to the external stimulus is not possible. On
the contrary, it is only by means of a systemic reading of the environment (the particular
phase of the cycle, the characteristics of the agents surrounding the firm, and the rules
governing the interactions) that the impact on micro decisions can be analyzed. In this
respect, Freeman’s historical perspective in particular, and the IS approach in general, can
shed light on the direction and intensity of innovation during the different phases of the
economic cycle.
In the case of institutional instability and how it can lead to shorter economic cycles (the
second part of my statement), the explanation is based on the fact that it increases the level
of uncertainty, shortens the horizon of investment decisions, narrows the number of
profitable innovative investments, and allows a wide range of moral hazard situations.
These types of environments create incentives to survive that are not based on innovation
(Fajnzylber, 1989) and allow firms to compete based on rent-seeking, enclave-type
economies and informal (even illegal) economic activities (Kosacoff and Ramos, 2006;
Bottazzi et al., 2010; Altimir and Beccaria, 1999).
From the point of view of demand, unstable environments work against sophisticated
tastes and preferences given the instability of incomes, with the consequent impact on the
level and characteristics of production. Of course, it also impacts the trust of institutions,
affecting the relationship individuals have with them. In this case, the clearest example is
the financial system and the extent to which banks serve as intermediaries between the
supply of savings and the demand for loans. Another example is confidence in the tax
system and people’s willingness to pay their taxes. If citizens do not trust public authorities
and how they administrate the public budget, they will have more incentive to avoid paying
taxes, with the consequent impact on the public budget and, of course, how the
government administrate it.
Chapter 1: From a “focusing device” to a matter of public policy
- 45 -
If a more systemic view is taken, unstable environments lead to the deterioration of
education, health, and security, which impact individuals’ freedom (to use Sen’s
terminology) and capabilities, which impact on the system’s process of competence
building. For example, any change in the organization of the educational infrastructure (a
change in the public authority in charge or a change in the curricula) will lead, in the short
term, to a decrease in the efficiency of the organizations affected, in that new routines have
to be set and internalized. If changes occur repeatedly, the loss of efficiency will be more
the rule than a short-term impact, and these institutions will be more focused on dealing
with change rather than improving how they teach (or how patients are treated, when it is
the health system facing changes in the rules of the game). The worst part of this situation
is that the impact of the reduced efficiency in the educational infrastructure is not only
verified through lower levels of capabilities but also in the work force’s future levels of
productivity and the consequent impact of this on incomes.
In other words, Myrdal’s cumulative causation also applies in negative terms, and therein
lies the link between instability and development. If development is about growth with
equality and this is reached by the sustained enhancement of capabilities and the process of
competence building, and if the existence of instability works against both of these things,
then it blocks development. As shall be seen in the remainder of this chapter, the
relationship between development and competence building has been one of the key
elements in the controversies about how to apply the NIS approach to developing
countries.
1.3.2. Development and the IS approach
A question that has motivated a large number of theoretical and empirical analyses is how
the IS approach can contribute to the identification of causes and consequences of
economic growth with equity. That is, how to explain the existence of differences and how
to promote convergence. In this sense, a great controversy within the approach is whether
special considerations should be made when applied to developing countries or not.
Edquist (2001), for example, has proposed the concept of Systems of Innovation for
Development (SID), which—according to the author—is a variant of the general IS
approach that emphasizes developmental features in order to improve its relevance and
usefulness. In this variant, diffusion is more important than technological development,
process innovations explain more than product innovations, and technological change
must be sought in traditional sectors rather than high-tech sectors, etc.
Chapter 1: From a “focusing device” to a matter of public policy
- 46 -
This redefinition of the approach is opposed to Lundvall’s and Nelson’s ideas, though for
different reasons. For Lundvall (1992), there are no two identical systems and therefore,
besides differentiating between developing and developed countries, distinctions should be
made between northern and southern countries, American and Asian countries, etc. For
Nelson and Dahlman (1991), the IS in developed countries are a benchmark for analyzing
developing countries: differences in the productive structure, in the state’s role, in the
dynamics of innovative firms, in the generation of knowledge, in the impact of capital good
suppliers, in the production of S&T, etc.
To some extent, Edquist’s approach contains, implicitly, Rostow’s concept of development
where becoming developed depends on going through a series of stages, which occur along
a linear path (Rostow, 1962). However, the misunderstandings of this explanation have
been widely demonstrated, even by his contemporary colleagues. In practical terms, the
development of Asian countries and the boom of large countries like Brazil, China, and
India show significantly different paths. In this sense, Freeman would point out that given
the global scenario (the Cold War world is different from the world during the 2008 Global
Financial Crisis), the overall dynamic of national institutions explains the different
performance of some nations, which are hardly comparable with the performance of others
(it is hard to believe that Brazil and Ivory Coast would ever follow similar development
paths).
From a theoretical point of view, Reinert’s (1996) contribution to the explanation of
development/underdevelopment is enlightening and has helped to shape the way the IS
approach is applied to less developed countries. This author emphasizes that the
productive structure poses limitations to development when technical progress occurs in
the classic form: lower prices, lower wages and, consequently, reduced welfare,
unemployment, exclusion, and deterioration of the income distribution (for example, the
introduction of bar codes in supermarkets reduced the required skills of checkout clerks
and therefore, their salaries). On the contrary, when technological progress takes place in a
collusive way—higher prices and wages—it generates the opposite effect and the nation
benefits through an increase in its incomes (to keep up in the “mobile race”, improved
skills are required every day, so higher salaries are paid).14 The specialization profile and the
14 Reinert (1996) uses the concept of “classical” forms of distribution of the gains from technological change to explain that in some cases technical change increases productivity levels, reduces the requirements for skilled labor, and leads to lower prices. This is the case observed especially in the traditional sector. There are sectors, however, where technological change creates
Chapter 1: From a “focusing device” to a matter of public policy
- 47 -
environmental characteristics are key elements for predicting the type of technological
change that will take place. Each new breakthrough, in the terms of Dosi (1982), represents
a new opportunity for entering the world market during the selection process, where
competition has not yet eroded prices and imitation has not standardized the product, so
most technological progress has the form of collusive improvements.
By definition, developing countries are characterized by productive structures with low
value added, located in markets with low income elasticity and lower wages. Therefore, the
variant proposed by Edquist openly opposes the dynamics Reinert describes, as Edquist’s
approach implies sustaining the growth path of the dominant (traditional) sectors, while in
Reinert’s explanation this would imply reproducing underdevelopment. That is to say, the
approach proposed by Reinert would argue that the SID variation is not only less relevant
to the IS approach, since it does not contribute new tools for understanding the
phenomenon, but also its acceptance would perpetuate underdevelopment.
In this sense, Nelson and Dahlman’s (1991) contribution in terms of regularities in
developing countries seems more accurate. In their comparative study of several countries,
the authors identified some common characteristics of developing countries, associated
with how technology is incorporated, the impact of economic history, and the importance
of absorptive capabilities (dissemination). Using these specificities, and given the evolution
of more developed countries, the authors propose the concept of catching-up but are
aware of the fact that technology is not freely appropriable (something that is absent from
Edquist’s work). On the contrary, leader nations seek to perpetuate their position by
creating barriers to diffusion, and not by new Schumpeterian-style innovations. Therefore,
closing the technology gap does not mean merely incorporating and adapting technology
but rather creating endogenous capabilities, supported by an environment capable of
overcoming entry barriers (intellectual property regime, macroeconomic stability, local
technology development, etc.).
A more radical position is that of Lundvall (1992), who claims that the IS approach is
actually a tool to identify specificities. This does not mean that comparative studies cannot
be carried out, but this should happen through systemic approximations and not by
comparing inventoried means. Lundvall et al. (2009b) discusses how the approach can be
entry barriers (patents, scale), which preserve the market from perfect competition. To sustain the barrier, more complex skills and higher investments are required. This is the case of the so called “dynamic” sectors and what Reinert named collusive technological change.
Chapter 1: From a “focusing device” to a matter of public policy
- 48 -
used as a tool for policy-making, distinguishing between the “mature” and the “emerging”
IS, a classification easily comparable to that proposed by Edquist and Nelson in the sense
that also highlights the passage from one stage to the other. However, the practical
implications of Lundvall et al.’s interpretation are different: they emphasize the importance
of creating basic organizational and technological skills (engineering, R&D, knowledge
management), whereas Edquist suggests that the diffusion of technologies is key to
development. Lundvall et al.’s explanation is similar to Nelson and Dahlman’s account of
firms, but they extend it to all institutions, especially those in the public sector, and their
ability to design, implement, and monitor policies. They also add that the determinants of
underdevelopment must be sought in systemic failures (as opposed to market failure), and
that the passage from an emergent to a mature system presents unique and distinct paths,
to the extent that that what constitutes a systemic failure in one country may not be one in
another. For the same reason, comparative studies cannot be abstracted from the socio-
economic conditions and the evolving structure of the analyzed IS. As a consequence, the
authors move away from the functional approach proposed by Edquist (2001), who argued
that a certain series of requirements where necessary to advance development, but also
from that of Nelson and Dahlman (1991), who assumed that a positive feature in one
country (e.g. high R&D expenditure) can be reproduced in another with the same impact,
allowing it to catch up.
1.3.3. The Latin American view
In what follows, the main Latin American contributions regarding the relationship between
the IS approach and its practical implications are presented with the objective of identifying
and analyzing some key findings that constitute the shared characteristics of the region and,
therefore, of Argentina. At the same time, these contributions are directly based on the IS
approach and constitute part of the new literature that has no’t yet been fully integrated
into the approach. The major contributions of Latin American scholars with regard to
three key questions will be reviewed, taking into account the characteristics of the region,
the purpose of this chapter, and the general objective of this thesis.
Responding to some extent to the controversy between Lundvall and Edquist, the first
question has to do with whether or not a specific approach for developing countries,
especially Latin American nations, is required. The second question, related to the
specificities raised by Nelson and Dahlman and close to Freeman’s historical analysis,
relates to how a particular history of instability led to a specific range of opportunities.
Chapter 1: From a “focusing device” to a matter of public policy
- 49 -
Finally, in connection with the practical implementation of the approach, the third question
deals with the tension between the theoretical postulates and their empirical application.
It is important to state that the recognition of the role of innovation and technological
change was not only a phenomenon for developed countries. There is a long tradition of
analyses about unequal growth, zero-sum games, and technological gaps among Latin
American scholars and policy makers. During the mid-twentieth century, similar ideas were
found in the schools of Latin American structuralism and dependentism, which understood
innovation as a determinant of growth and development, and capitalism as a process of
disequilibrium and imperfect competition.15
With this background, the IS approach reached the region in a context of generalized
liberalization and deregulation (the Washington Consensus) and constituted a powerful
tool to explain the logic behind the national plans and programs to promote science,
technology, and innovation. Moreover, in some countries, the promotion of the NIS was
explicitly recognized as a public policy objective.16 The 1990s witnessed the passage from
supply- to demand-promotion policies, on the assumption that the obligation to cooperate
as a condition of accessing competitive funds would solve decades of disconnection
between knowledge supply and demand. This context framed the academic work of the
time, which has contributed to the regional view of the IS approach.
1.3.3.1. A specific approach vs. a specific approximation to development
In relation to whether or not a particular approach is required for Latin American
countries, local contributions are closer to Lundvall’s position than to Edquist’s. Regional
scholars have argued that a specific IS approach for developing countries is not possible,
given the heterogeneity of situations both between and within continents (for example,
underdevelopment in southern Africa is significantly different from that in Latin America,
but the situation in Honduras also differs from the one in Uruguay). As such, each type of
15 The Prebisch-Singer thesis on the secular deterioration of the terms of trade was written during the 1940s and gave support to the process of industrialization by import substitution in the middle of the twentieth century (Prebisch, 1950; Singer, 1950). The description of Sábato’s Triangle was first published in 1970 and helped to understand technological dynamics as part of a complex system which included firms, government. S&T, and education centers (Sábato and Botana, 1970). Herrera’s ideas on the impact of environmental incentives (which he called implicit policy) and their relationships with explicit policies can explain many of the policy failures of the twentieth century (Herrera, 1971). Fernando Cardoso, Celso Furtado, Andre Gunder Frank and many other scholars from the region argued that industrialization was the only mechanism for development and that the generation and adaptation of technology the key activities for it. 16 See for instance Porta et al. (2011) for the cases of Argentina, Brazil, Paraguay, and Uruguay, Dutrenit et al. (2010) for the case of Mexico.
Chapter 1: From a “focusing device” to a matter of public policy
- 50 -
situation will demand a specific approach, what is exactly what Lundvall meant when he
defined the IS concept as a focusing device.
Despite the consensus on the methodological nature and implications of the approach,
while for Lundvall it is an ex post concept, its application to the Latin American case has
been based on an ex-ante interpretation and analyses have focused on explaining how the
systemic dynamics stated in the theory are not confirmed in practice (Dutrenit, 2009;
Arocena and Sutz 1999). Thus, while in its genesis the concept is a formulation of the
regularities observed in developed countries, in Latin America it spread as a tool to explain
the lack of systemic dynamics in the multiple dimensions of the innovation process. This
same ex-ante application led to an emphasis on the normative dimension of the approach,
transforming the concept of IS into a policy subject (Arocena and Sutz, 1999).
The “view from the south,” in the terms of Arocena and Sutz (1999), or the “South
American framework,” in the terms of Cassiolato and Lastre (2002), emerged as a way to
emphasize key elements that should be considered in any analysis of the countries in the
region, in relation to its history, its current structure, and its hierarchical position in the
world. However, this alternative is not similar to Edquist’s SID. On the contrary, these
analyses reject the idea of stages of development. Instead, they argue that the challenges
Latin America faces in the new economy are not the same as those faced by Europe,
America, or Japan and are also different to those faced by other Asian or African countries.
Thinking that these countries could follow the same steps is not only wrong but dangerous.
1.3.3.2. Causes and consequences of development
As was mentioned before, Nelson and Dahlman outline some specificities of developing
countries that determine their worse innovative performance. However, these specificities
also shed light on how to transition from a stage of underdevelopment to one of virtuous
growth. Like Edquist’s proposals, this interpretation of how development should be
reached could lead to the perpetuation of the technology gap. If in developing countries
process innovations are more important than product ones, then promoting the
incorporation of machinery and equipment will perpetuate this situation unless the
technological frontier is reached and moved. Moreover, although the authors recognize the
importance of endogenously generated knowledge, they do so in relation to these same
identified specificities (continuing the example, this would imply the development of
endogenous capabilities in process improvement).
Chapter 1: From a “focusing device” to a matter of public policy
- 51 -
These criticisms from Latin American authors have to do with the fact that what the
founding fathers of the IS proposed as causes of underdevelopment are actually
consequences. Indeed, the region is characterized by a production structure that is biased
towards the industrialization of primary products (commodities) and assembly
(maquiladora), with low knowledge content, weak production linkages, and few dynamic
firms which are not enough to pull an overall increase in productivity (Cimoli, 2005;
Dutrenit and Katz, 2005; Lugones and Suárez, 2006; Porta, 2006; Katz, 2007; Ocampo
2012). Therefore, the higher weight of process innovations, the importance of diffusion
processes over the endogenous creation of technology, and economic instability are
actually consequences of the productive dynamic, which is what explains the lower relative
levels of development.
In the same sense as Reinert’s explanations, but taking into account the existence of what
Lundvall called systemic failures and Freeman’s historical determinants, the development
of Latin American countries depends on their ability to break the vicious cycle caused by a
selection environment that works against innovation processes (historical and structural
characteristics) and existing innovation processes that fail to alter the selection
environment (few virtuous companies). This requires not only structural change but the co-
evolution of institutions that support it (Dutrenit, 2009; Dutrenit et al., 2010). Once again,
the specificities of Nelson and Dahlman and the “pre-development” features raised by
Edquist are actually a result of an underdeveloped productive structure. These postulates,
therefore, are useful for explaining underdevelopment but not for providing elements to
modify it.
1.3.3.3. Theory and practice
Another criticism that the IS approach has received is regarding the distance between
theoretical postulates and their practical application. At the aggregate level, there are two
issues: the impact of the identified specificities and the systemic nature of innovation.
Because of regional specificities (specialization, links, and a history of disequilibrium), the
innovation process in the region is characterized by a low level of innovation investments
towards technological and organizational improvements, focused on the acquisition of
capital goods, a weak network of connections, and a large number of firms with positive
innovation results but with reduced scope for the innovations—Nelson and Dahlman’s
specificities (Anlló, et al., 2007; Dutrenit and Katz, 2005). These characteristics mean that
Chapter 1: From a “focusing device” to a matter of public policy
- 52 -
although a general methodological approach for all countries is vital, regional and national
approximations are also important. The former is required to address the issue of the
technological gap and best practices, the latter two to understand, characterize, and
monitor the evolution of the specificities explained above and to try to identify policy
recommendations (Lugones and Suarez, 2006; 2010).
The second issue has to do with the systemic nature of the innovation process and how the
excessive emphasis on comparative measurements and objective indicators has worked
against the characterization of the systemic nature of the innovation process. In an
inventory-type analysis, Latin American countries show the same institutions and the same
frequency of linkages as developed countries. However, the innovative dynamic, the role of
intangible institutions, and idiosyncratic characteristics determine disarticulated systems,
with clear differences between the processes that emerge in Europe or the United States
and the ones that emerge in Latin American countries.
In this sense, criticism of the empirical application of the approach relates to the low
attention paid to connectivity capabilities and the co-evolving nature of the system. The
former is associated with the study of linkages in terms of two-way flows of information
and knowledge and its impact on the innovative skills (Yoguel and Robert, 2010). The latter
is concerned with interdependence and the evolutionary features of the system (Dutrenit et
al., 2010). In this sense, inventory-type approaches reduce the attribute of complexity to the
construction of composite indicators, without the inclusion of feedbacks, accumulation,
and dynamic trajectories.
For the last twenty years, analyses based on international comparisons have led to the same
conclusions over and over again: developing countries have to invest more in R&D and
increase their levels of qualified human resources, which is basically the main
recommendation of studies based on the concept of catching-up.17 These studies show
obvious consequences, they are useful situations for descriptive purposes but insufficient in
identifying key policy criteria in the sense of the ex-ante application of the approach. The
problem with this static analysis is the danger of confusing means with ends and the fact
that each agent will respond differently given its environment and history. Consequently,
17 Given its importance among the initial theoretical contributions to the ISs approach, only Nelson and Dahlman’s comparative study has been referred to. There are, however, a significant number of cross-country comparisons such as the ones presented in Alburquerque (1999), Godinho et al. (2004), Anlló and Suárez (2008), CEPAL (2008); Cimoli (2005), and the ones discussed in Suárez (2006); all of which reach similar conclusions.
Chapter 1: From a “focusing device” to a matter of public policy
- 53 -
the actual possibility of reproducing successful cases is reduced, if not impossible, given the
number of variables that should be controlled (or fostered). In other words, the usefulness
of the approach as a focusing device disappears when it is converted into a methodological
tool based on comparative statistics.
If innovation is more than R&D and if the low level of investment is a consequence (and
not a cause) of underdevelopment, then these conclusions are tautological and cannot shed
light on how to change the technological trajectory of developing countries. Comparisons
of developed versus developing countries should be the starting point of any research, to
the extent that they can characterize the technological and developmental gap. However,
they are not enough to translate the approach into a research tool capable of determining
how to close the gap.
To conclude, the application of the approach to Latin American countries, and the
theoretical and methodological contributions derived from this application have biased the
approach into a static one, where inputs and outputs can be described and measured with a
great level of precision, but the process of innovation at the national level is still a black
box. Opening that black box implies understanding how similar inputs lead to different
results and why the same formalized institutions lead to different rules of the game. In
other terms, it implies understanding the dynamic, cumulative, and path dependence
characteristics of the innovation process at any level.
1.4. Preliminary conclusions
In this chapter, the IS approach has been presented in an attempt to characterize its
complexity and diversity. In a very stylized definition, it is a set of contributions that may
be more or less articulated but that have a clear unifying goal: to explain the dynamics of
growth and development on the basis of technological change and innovation. Based on
this objective, explanations and contributions are numerous, though mostly
complementary. The lack of clear boundaries, a unified body of theory, and fundamental
laws are perhaps the main advantages of the approach and a proof of its potential as a
methodological tool to explain both why a country, region, or industry is located in a
certain place in the development ranking and what process led to its being there. This lack
of boundaries is also its main limitation as long as almost everything can be included as part
of the NIS.
Chapter 1: From a “focusing device” to a matter of public policy
- 54 -
During the last two decades, the approach has led to the diffusion of several empirical
analyses that have allowed scholars and policy makers to come a little closer to an
understanding of development and an identification of key criteria for the design,
implementation, and monitoring of public policies. However, it is also in this regard that
the approach has received its main criticisms, some due to its translation into indicators
and variables, others because of its automatic application to developing countries. As such,
serious challenges still remain, in particular, the distance between theory and practice.
There is also a set of issues associated with the advantages of this approach, which explains
the wide diffusion of its main arguments. The first of these has to do with the apparent
dichotomy between its ex-post and ex-ante application. The tension between the
descriptive and prescriptive usefulness of the approach revealed the dual intention of its
application to concrete cases, which in turn makes this dichotomy no more than a
theoretical appreciation. Indeed, to use the approach as a policy tool, a benchmark is
required. Consequently, it is applied as an ex-post concept in developed countries and then
applied as an ex-ante definition in developing ones, in order to analyze the gap that should
be closed if a catch-up process is to take place.
Another useful issue has to do with the recognition of the role of the public sector and the
world scenario in the quest for development. In this regard, despite the criticisms about its
translation into practical studies and its limitations when trying to reproduce specific
scenarios, the IS approach has proved a powerful methodological tool for identifying the
challenges faced by developing countries in closing the technological gap and it is a
vindication of the duty of policy makers to foster development in an increasingly globalized
world. At the same time, the approach recognizes that each country occupies a particular
place on the world stage, that sub-national approaches complement the approach, and that
global capitalist reality marks the political, social, and economic limitations to the pursuit of
development.
In relation to Latin America, regardless of the degree of communion with the IS approach,
the discussions that have been taking place since the publication of Lundvall’s (1992) book,
resulted in a Latin American network of scholars with a long tradition in the study of
technological change and development.18 As a consequence, this approach has triggered an
18 For instance, the annual Globelics Conferences, the biannual conferences of the Latin American
Asociation of Technological Management (Asociación Latinoamericana de la Gestión Tecnológica, ALTEC), the BRICS project, the activities of the Iberoamerican Network of Science and Technology Indicators (Red Iberoamericana de Indicadores de Ciencia y Tecnología), the Schumpeterian Society, etc.
Chapter 1: From a “focusing device” to a matter of public policy
- 55 -
extended research agenda which has impacted national public policies and social awareness
in the region.
Regarding the theoretical aspects of the approach, regional contributions agree that
innovation is a key to development, which emerges from a geographically and historically
rooted process with a systemic and interactive nature. The trajectory of instabilities in the
region, the persistence of average productivity levels below international standards, and the
existence of strong social inequalities demand that the issue of science, technology, and
innovation be addressed, taking into consideration the heterogeneity of situations, the non-
linearity of processes, and, especially, the fact that successful public policies in developed
countries could impact Latin American countries differently. As such, the general approach
allows regional specificities to be identified.
Finally, it is worth reflecting on the relationship between the deterministic fatalism and the
naive optimism of some particular approaches. No author would disagree with the need for
long-run growth, the importance of adding value, or the need to foster development. In
fact, this is not a distinctive feature of the approach but it is something that economic
theory has preached for decades. However, the focus of the IS approach and the strong
emphasis on national realities has given rise to arguments against the idea that policies can
be applied in different places with the same results, that the economic dynamic is
determined exclusively by individual decisions, and that technological catching-up will
happen by mimicking the productive development of northern countries. Arguments like
these have led to conclusions such as that the lack of investment in R&D is the cause (and
not the consequence) of underdevelopment, that a better intellectual property system
would allow the development of so-called high-tech sectors, and that despite the systematic
failure to add value to natural resources, development can now occur from the inclusion in
high-value niches with virtuous linkages (and multiple entry barriers).
In the face of analyses like these, IS studies are a theoretical and practical alternative that
accept that economic reality is much more complex, that the systemic nature of innovation
makes it impossible to reproduce scenarios, and that the path to development depends on
the confluence of public and private intentions in a context of asymmetric relationships. As
a result, it is in many ways a realistic approach that explains the past and identifies possible
futures. It is precisely this understanding of the approach that will be applied to the
Argentinean case in the following chapter in order to analyze the process of learning,
competence-building and innovation within an unstable environment.
- 56 -
Chapter 2: A historical analysis of the process
of competence building in an unstable
system
Introduction
The translation of the IS approach from a focusing device to a concrete methodological
tool demands the design of a model capable of simplifying complex reality to specific
dimensions. In this sense, this chapter has a double objective. It aims firstly to propose a
scheme to analyze NISs, and secondly, to apply the proposed scheme to the Argentinean
case in order to study the impact of an unstable environment on firms’ competence
building and innovation process.
The selection of the national level responds, on the one hand, to the fact that this level
allows the different dimensions which during specific periods have shaped the system
dynamic to be integrated. On the other hand, and conversely, the selection of the national
level will allow an understanding to be reached of how the aggregate behaviors of different
organizations led to a particular configuration of the environment that is more or less
prone to innovation. In other words, it allows the bidirectional causality between
organizations and environment to be studied, which in turn will contribute to
understanding the relationship between the national system and micro-heterogeneity.
Given the theoretical framework and the aim of this chapter, it is worth mentioning that it
does not aim to present a complete analysis of Argentina’s history but instead to review the
major events that have impacted the selection environment in which firms have operated in
every particular period, including the role that innovation has had in the firms’
competiveness.
The questions guiding this chapter are related to establishing the key elements of the
environment that have impacted (and impact) the innovative behavior of Argentinean firms
and determine the set of incentives to innovate. More precisely, it seeks to identify the
elements that explain the set of (dis)incentives to compete in each historical moment. Of
course, this chapter also aims to test the usefulness of the NIS approach to explain the
innovation process in unstable countries.
Chapter 2: A historical analysis of the process of competence building in an unstable system
- 57 -
To answer these questions, this chapter is structured as follows. Section 2.1 presents the
methodological approach and discusses some contributions thatallow the scope and
implications of the main concepts of the NIS approach to be defined and translated into
practice. Section 2.2 summarizes the main events that have impacted Argentina’s
productive dynamic over the last century. Section 3.3 arises from the integration of the
methodological framework and the dynamics of the Argentinean innovation system during
the last century. Finally, some preliminary conclusions are provided.
2.1. Methodological discussion
The Republic of Argentina is a Spanish-speaking country with Latin roots and 40 million
inhabitants. Located at the southern end of South America, it is a producer and exporter of
natural resource-based goods. During the two-hundred year history of the modern state,
the constant has been the presence of strong economic, social, and political instability,
cyclically repeated over the last century. Argentina’s productive structure is less competitive
than Brazil but more so than Colombia, Chile, and Venezuela. Its geographical size
distinguishes it from the smaller countries in the region such as Uruguay, Peru, and
Ecuador. In terms of development, the early consolidation of labor unions and social
protection laws, a universal health system, and free public education have made Argentina
one of the Latin American countries with the best levels of income distribution and human
development indicators. However, recurrent crises have created zones of poverty and
exclusion, which have not disappeared even after more than a decade of continuous
growth.
Argentina’s economic history since the late 19th century can be divided into four distinct
periods: the agro-export model (1880–1930), industrialization by import substitution
(1930–1976), liberalization and deregulation (mid-1970s– 2002 devaluation), and the new
process of growth that characterizes the country today, which we will refer to as “the
current model.”. In each period, the role of the public sector, productive incentives, and
the relationship between Argentina and the rest of the world have marked the direction of
technological change and helped to shape the current specialization pattern.
In light of the NIS approach, Argentina’s NIS has some structural features which are the
result of path dependence processes that can only be understood from a long-term
perspective. In this sense, and as I shall prove, although the more obvious and palpable
impact arises from the dynamic of the system during the last 20 years, there are other
Chapter 2: A historical analysis of the process of competence building in an unstable system
- 58 -
features that can only be understood if one analyzes the history of the last century: the
predominant specialization pattern, the social, political, and economic structure, and
Argentina’s particular idiosyncrasy.
The observable features of an NIS are assumed to be the result of an evolutionary process
from which competencies are created and built, a process which includes not only firms
and their environment but also the intensity and direction of the interactions between
them. As a consequence, this analysis then takes up Freeman’s (1995) historical perspective
and Lundvall’s (2009) broad interpretation of environment. The proposed framework is
based on the assumption that the narrow definition of the system is incomplete and only
accounts for the situation of the knowledge infrastructure but not for the impact of the
context on the innovation process. In other words, the knowledge infrastructure will be
explicitly differentiated from those other organizations that have fostered or driven
innovation at each particular moment.
The methodology for the study of Argentina’s NIS is the presentation and analysis of
organizations, institutions, and linkages which have determined the dynamics of
competence building and growth, paying special attention to structural changes that
transcended the particular economic model and the challenges that had to be addressed in
the following periods. The general assumption is that the relevant dimensions for
explaining innovation could change from one economic model to the next, and thus there
is no a priori selection of institutions or organizations that can be repeatedly analyzed in
each period. Special attention will be paid to the role of institutions in the sense of Johnson
(1992), in an attempt to test the hypothesis about institutional delay versus Johnson’s
understanding of institutions as facilitators of change.
As the objective of this thesis is the innovative behavior of manufacturing firms, this type
of firm is placed at the center of the analysis and each of the analyzed dimensions will be
linked to the impact on firms. In this sense, the idea is to examine in greater depth the
historical determinants that led to the stylized facts of the productive and innovative
dynamic presented by Dutrenit and Katz (2005) and by Lugones and Suárez (2010).
Figure 2.1 schematizes the conceptual approach. According to this figure, the innovation
system is made of six types of organizations, which can be grouped into three major
categories: supply and demand, the infrastructure of science, technology, and education,
and the government.
Chapter 2: A historical analysis of the process of competence building in an unstable system
- 59 -
Figure 2.1: Conceptual scheme of the National Innovation System
The “supply and demand” group is made up of three types of organizations: firms, which
account for the productive structure at the most aggregate level (primary, secondary, and
tertiary sectors) and constitute the center of this analysis; the supply chain, with emphasis
on the labor market, the financial market, and the suppliers themselves; and demand, which
will be analyzed in terms of income levels and preferences.
Within this framework, innovation is assumed to happen within firms (innovations are
taken into the market by firms), although the innovation process results from the
interaction between the firm and the rest of the system: the firm’s suppliers and customers
in the sense of Lundvall (1992), the knowledge infrastructure in the terms of Sábato and
Botana (1970), or even the public sector, as a result of access to public policies.
The science, technology, and education infrastructure includes laboratories and public and
private education facilities and universities. The former are assumed to be more associated
with knowledge production and the latter with the training of human resources.
The public sector includes all government areas associated with the design and
implementation of public policies. State-owned enterprises are assumed to be part of the
supply and demand sector. The macroeconomic dynamic in its wider sense is assumed to
be the result of public policies and therefore is included as part of the analysis of this
sector.
The arrows in the figure represent the interactions between supply and demand, the
knowledge infrastructure, and the public sector. These arrows account for both the direct
Supply Chain
Inputs and raw materials
Capital goods
Labor Supply and Demand
Science, Technology and Education Infrastructure
Science and Technology System
Schools and Universities
Demand
Public Sector
Firms
Primary S.
Secondary S.
Terciary S.
National Innovation System
Chapter 2: A historical analysis of the process of competence building in an unstable system
- 60 -
relationships associated with the innovative process within the firm and the indirect
interactions that impact the qualification of human resources (the interaction between the
knowledge infrastructure and the labor market); the improvement of inputs by means of
scientific and technological activities (interactions between the knowledge infrastructure
and the supply chain); consumer purchasing power (the interactions between the labor
market and demand); etc.
Direct and indirect interactions account for the exchange of information, goods, and
services, and that is why they are assumed to be bidirectional. Of course, this does not
mean that the flow in one direction is assumed to be equal in quality and quantity to the
reverse flow (Yoguel and Robert, 2010). Special attention will be paid to the co-evolution
of some organizations in order to test the hypothesis of mismatches between the direction
and intensity of the competence building among the science, technology, and educational
infrastructure and the productive sector (Dutrenit et al., 2010; Dutrenit and Puchet, 2011).
One element not included in the figure is the flow of information, goods, and services
between the system (and its components) and the rest of the world, as proposed by
Cassiolato and Lastre (2002). These authors refer to a country’s position within the
dynamics of global capitalism, the international incentives that lead firms to export or
import and scientists to interact with their peers, the diplomatic relations between the
country’s government and other governments, and those international agreements that
shape—and limit—national policies. In this respect, this dimension will be referred to and
characterized when relevant to explaining the innovative dynamic of the system.
In short, organizations, relationships, and historical processes are the axes of the analysis
presented below, which aims to identify the key dimensions of the Argentinean innovation
system which explains innovative behavior at the firm level and the system’s competence
building process.
2.2. An overview of the Argentinean NIS
2.2.1. The agro-export model (1880–1930)
The period stretching from Argentina’s independence from the Spanish crown in 1816 to
the beginning of the agro-export model in 1880 is known as the period of institution
building, during which the foundations of today’s federal state as were established. During
this period, Argentina began to emerge as a global food exporter, with a high level of
Chapter 2: A historical analysis of the process of competence building in an unstable system
- 61 -
specialization in grains. At the same time, livestock farming grew, but was oriented towards
the satisfaction of local demand—basically leathers and jerky and, to a much lesser extent,
meat. Manufacture was based predominantly in small manual workshops that aimed to
satisfy the needs of city-dwellers, particularly the growing settlement around the harbor on
the Río de la Plata that is now Buenos Aires, Argentina’s capital.
In 1880, three major innovations arrived at Buenos Aires: communication though the
telegraph19, the revolution in maritime transport, and refrigeration. The country’s economic
policy during this period was aligned with the boom of the global free-market, and
interventions were related to these innovations and the need to consolidate national
integration: public investments in communication infrastructure, the promotion of
immigration, and attraction of foreign direct investment (FDI).20 As a result, between 1895
and 1930, Argentina’s population grew from 3.8 to 11.2 million, mainly due to the arrival of
farmers from Italy and Spain (Bunge, 1940). FDI allowed the rail network to be
consolidated (using British capital) and large refrigeration facilities (from the United States)
to be installed, thus rendering Argentina capable of responding to the increasing worldwide
demand for fresh meat (Cortes Conde, 1963).
The impact of these innovations and promotion policies is undeniable, as is the fact that
the fertility of the land and climate conditions led to high levels of cost competitiveness. By
1930, Argentina was an established producer and exporter of grains and meat, with higher
productivity levels than its main international competitors and a clear supremacy over the
other two large food producers, the United States and Canada.
Argentina’s rise to prominence as a global food and primary-sector producer during 1880–
1930 boosted both a significant increase in GDP and in technological progress: between
1910 and 1930, domestic production increased by 130% and the production of primary
products reached 60% of the total production value (Bunge, 1940).
Argentina’s manufacturing industry began to develop pari pasu the GDP growth but, unlike
the primary sector, it did so by looking to the domestic market, where demand was rapidly
19 Although the telegraph had been installed in some railway stations, and strategic positions in the countryside had been connected for military-political reasons, it was not until the presidency of Domingo Faustino Sarmiento (1868–1874) that the installation of the telegraph following railway infrastructure became public policy. 20 Although tariffs to foreign trade were higher than those observed in countries with similar development levels, they had to do more with the public need for funding than with an explicit policy of export promotion or national industrialization (Diaz Alejandro, 1967).
Chapter 2: A historical analysis of the process of competence building in an unstable system
- 62 -
expanding due to the consolidation of the country’s major cities and population growth.21
In this sense, the production of goods with natural barriers to import, such as cement and
bricks, and the production of basic consumption goods associated with local tastes and
preferences (food, textiles, and chemicals)22 were the sectors with the highest participation
in the total industrial product. Surprisingly, despite the strong impetus towards the
incorporation of technology in the primary sector, this sector turned abroad to seek
demand, the few linkages that were generated with local firms were basically with the infant
metalworking industry, which would eventually specialize in railway parts.23
The service sector followed a similar path. During this period, financial and trade services
accounted for almost all of the sector’s activity, which were provided by British firms,
Argentina’s main trading partner and provider of freight, insurance, and financing services
through the advancement of crops. Local businesses, in turn, concentrated their activities
on the provision of personal services associated with the development of cities, while the
public sector was in charge of the provision of urban services.
Regarding demand, unlike what was observed in other countries in the region, a
homogeneous labor market with low levels of informality led to high purchasing power.
The cause and consequence of the homogeneous labor market was the universalization of
the primary education, something that distinguished (and continues to distinguish)
Argentina from its regional peers, and the flowering of the liberal professions, which was
possible given the consolidation of public universities.24
However, the precarious situation of rural workers, the stagnation of job creation, the
growing gap between the richest and the poorest classes, and the housing problems typical
of growing cities started to put pressure on the model and demand a change in public
policy (Diaz Alejandro, 1975).
The Great Depression, World War II, the rise of Keynes’s economic perspective, and
protectionist policies in North America and Europe were another set of factors that
impacted the model. They drove the first policies promoting “strategic” industries from
21 Between 1885 and 1930, Argentina’s urban population grew from 1.6 to 7.6 million (Bunge, 1940). 22 In 1914, 54% of the industrial capital of Buenos Aires was accounted for by food manufacturer facilities, 2.2% by the textile industry, 6% by metalworking shops, and 2.2% by the chemical sector (Cortes Conde, 1963). 23 This demand for parts and pieces was the start of the metalworking industry and formed the basis for its promotion from 1945 up to the 1970s. Such was the impact of this demand that for several decades, these workshops were a key place of human resource training. 24 Between 1914 and 1938, the illiteracy rate of the total population dropped from 35% to 12%, in contrast with levels higher than 50% in other countries in the region and 40% in the Iberian Peninsula (Bunge, 1940).
Chapter 2: A historical analysis of the process of competence building in an unstable system
- 63 -
1930 on, together with the start of the first military dictatorship that same year (Katz and
Kosacoff, 1998). A few years later, these first policies led to a model of industrialization
that would later become known as the model of industrialization by import substitution
(industrialización por sustitución de importaciones, ISI) and the birth of the modern manufacturing
industry (Villanueva, 1972).
2.2.2. Industrialization by import substitution (ISI) (1930–1976)
2.2.2.1. More than an industrialization policy
In 1930 the manufacturing sector accounted for 20% of Argentina’s GDP, and four
productive activities showed strong potential for industrialization: metalworking, textile,
chemical, and food industries, which together accounted for 60% of manufacturing (Lopez,
2003). These sectors had been the main destination of domestic and foreign investments
during the previous decade had begun to incorporate technology before the world crisis,
and had a large margin for productivity improvements based on incremental innovations
(Villanueva, 1972; Katz and Kosacoff, 1998). The agricultural sector, meanwhile, had
reached the limits of extensive growth and faced declining foreign demand, due to the
international context. The government, which faced distributive pressures and employment
problems, had a foreign surplus and a strong industrialist ideology associated with the
popular imprint of Peronism.25
In this way, after 1930, industrialization became a state policy and the public sector
progressively increased its interventions to promote the development of the manufacturing
sector and to create institutions to support it. The first Argentinean wave of
industrialization (later called “easy ISI”), was based on direct public investments in the
provision of public services (energy, telecommunications, and fuels), the imposition of
tariff and extra-tariff barriers to external trade, a strong credit policy, and strong control
over the exchange rate.
Despite Argentina’s relative success with “easy” industrialization,26 the lack of an explicit
policy of coordination between the primary and the secondary sectors led to the birth of a
manufacturing industry with no articulation with the agricultural sector, with the exception
of large-scale grain producers who opted for downstream integration strategies. Thus, in
25 Peronism is a political and social movement named after Juan Domingo Perón, three times president of
Argentina, whose administrations were based on the ideals of industrial development with social justice and inclusion. See Cortes Conde (2005) and Canitrot (1975). 26 In 1965, the manufacturing industry accounted for 48% of the national value added (WB, 2012).
Chapter 2: A historical analysis of the process of competence building in an unstable system
- 64 -
spite of the strong demand for the mechanization of the primary sector, manufacturing was
unable to respond with specialized capital goods, which, added to the lack of export
incentives (among other things, because of effective protection27), determined a
manufacturing structure that was reduced in scale and was not integrated with Argentina’s
largest productive sector.
In theoretical terms, as “easy ISI” progresses, investments in the creation of infrastructure,
skills, and facilities should naturally take place in order to advance along the production
chain: from final goods to more heavy industry, and from simple technological inputs
towards more complex ones (Hirschman, 1968). However, there was never enough
strength within the manufacturing sector for this “natural” process to take place, and the
passage to heavy industry demanded, once again, state intervention.
In 1950, the government faced this new request for intervention with a mix of public direct
investment and the promotion of FDI. Public investment was concentrated in the
production of widely used inputs, and steel and electrical plants were installed. FDI,
meanwhile, was channeled into oil and automotive activities, and world-class multinational
corporations arrived in Argentina (Ford, Kaiser, Standard Oil, and Sevel, among others).
With regard to the local adaptation of technology, the state’s response to the new
challenges of industrialization led to the creation of what are now the main institutions of
the country’s S&T system (Bisang, 1995), namely: the National Council of Scientific and
Technical Research (CONICET) in 1958; the National Institute of Agricultural Technology
(INTA) in 1956; the National Institute of Industrial Technology (INTI) in 1957; and the
National Commission of Atomic Energy (CNEA) in 1950. These were influenced by the
linear model of innovation (that is, they distinguished between science, technology, and
innovation) and Vannevar Bush’s (1945) policy recommendations, and their establishment
was connected to the existence of a group of scientists and technologists with influence at
the government decision-making levels. Despite the undeniable importance of these
institutions, the way they were funded, their dependence on different ministries, and the
lack of an inter-institutional body to coordinate the whole system led to the consolidation
of isolated institutions that were only linked to their funders.
27 In order to promote the development of industries of widespread use, the promotion scheme established
tariffs on imported raw materials and intermediate goods, which meant that these rates were incorporated into the cost of final goods, but without a counterpart policy promoting exports, decreasing their price competitiveness.
Chapter 2: A historical analysis of the process of competence building in an unstable system
- 65 -
Public universities also benefited from public investments in technology, also under the
influence of the logic of the linear model. Consequently, Argentina’s whole education
system evolved without any coordination with the country’s productive needs, was oriented
almost exclusively towards teaching, and was based on degree programs leading to the
liberal professions.28 The creation of the National Technological University (1959) was
perhaps the only effort to create linkages between the manufacturing sector and higher
education, although its scope was not enough to motivate a change in the dynamics of
knowledge production and human resource training.
Finally, despite all the criticisms this period has received29, there are some areas of general
consensus about the structural changes that improved the dynamics of growth and
industrial development. The consolidation of labor rights, the improvement in income
distribution, and the universalization of health-care, education, and social security, which in
turn contributed to the improvement of general skills, are all hallmarks of this period. The
protection of the metalworking industry, the development of the food industry, and the
promotion of chemical research led to the foundation of the industrial sectors that
predominate nowadays and explain current employment dynamics. Peronism, as a social
and political movement, contributed to the consolidation of a society aware of the
importance of equality, income distribution, and social justice, which has managed to
overcome successive military dictatorships and repeated attempts to privatize health-care
and education. However, key problems cannot be ignored: these included the lack of
productive scale, the exclusive attention to the domestic market, the establishment of a
high inflation regime, the fiscal crisis, and the lack or articulation of key organizations,
which would impact technological dynamics for several decades.
2.2.3. The Washington Consensus (1976-2002)
2.2.3.1. The return to liberalism
The coup d’état of 1976 brought to power a new military dictatorship, which unleashed the
most profound violations of human, constitutional, and civil rights in Argentina’s history,
while also setting off a process of liberalization and deregulation. This change in economic
policy was intended to eliminate the “vices” of industrialization and push the economy
28 In 1930, of the 14.17 graduates per 10,000 inhabitants, 2.25 were engineers. In 1960, this relationship had increased to 37 graduates per 10,000 inhabitants, 8.8 of whom were engineers (López, 2003). 29 See, for instance, Azpiazu and Notcheff (1995) versus Katz and Kosacoff (1998).
Chapter 2: A historical analysis of the process of competence building in an unstable system
- 66 -
towards international technological levels (Canitrot, 1980). Throughout this neoliberal
period, the foreign and fiscal deficits set the limits of public policies and state interventions.
During the 1970s, the problem of fiscal and external deficits was tackled from three angles:
the reduction of the state, the boosting of comparative advantages (agriculture), and the
reform of the financial system (Ferrer, 2004). During its early years, the plan seemed to
work. However, it would be inappropriate to attribute economic success only to the
implemented measures. In the late 1970s and early 1980s, key economic sectors expanded
their activities in Argentina as in the rest of the world: soybeans, natural gas fields, the
fishing industry, and steel. Some expanded because of the maturing of investments of the
previous decades, others because of the new international context. This change in the
production structure affected several sectors, some of them on the verge of extinction,
especially the manufacturing of capital goods.30
During the first half of 1980, after a brief period of economic growth, tensions began to
rise again. Despite the downsizing of the state, the sustainment of labor levels had led to
oversized public investments—mainly in infrastructure—and then to fiscal crisis and the
birth of Argentina’s external debt. The change in international conditions (particularly the
reduction in the availability of foreign credit) and the loss of credibility in the government’s
exchange rate policy led to a new period of crisis and inflation. The return to democracy in
1983 was not able to change the economic situation that had begun during the dictatorship,
despite several failed efforts to stabilize the economy (including a new currency) and to
give momentum to the industrial and scientific policies (key S&T institutions were
reformed towards a more democratic organization). The results of this period include the
debt crisis of 198231, hyperinflation in 1989 (5,000% annual inflation rate), and a rising
unemployment rate (close to 10%).
2.2.3.2. The 1990s and structural reforms
Hyperinflation and the snowballing economic problems that followed the return to
democracy were partly responsible for President Raúl Alfonsín resigning six months before
completing his term in office. His successor Carlos Menem was elected in 1989, and two
years later a new solution was sought for the deficit problem. The set of measures
30 Between 1970 and 1987, the value added of the industrial sector had drop from 42% to 37% of GDP (WB,
2012) and the level of occupied workers had decreased to 26.6% (Katz and Kosacoff, 1998). 31 The debt crises of the 1980s were a widespread phenomenon all over Latin America, as were the liberal policies that had such a negative impact that the 1980s are known as the lost decade of Latin America.
Chapter 2: A historical analysis of the process of competence building in an unstable system
- 67 -
implemented by the Menem administration responded to the recommendations of the
Washington Consensus. Like the previous stabilization plan, its three pillars were the
downsizing of the public sector, the fostering of comparative advantage, and the reform of
the financial system (Kosacoff and Heymann, 2000). However, the ways these were
manifested in the 1990s were far more profound than the measures of the previous
decades.
To close the fiscal gap, a new wave of privatization was implemented, encompassing many
public firms and the pension system.32 To close the productive gap, the economy was
opened to international competition, external markets were deregulated, sectoral
promotion was renegotiated—especially for the automotive industry—and the labor
market was reformed towards a more “flexible” regime.
Inflation was fought by means of the deregulation of the exchange market and the
implementation of a fixed convertibility between a new national currency and the U.S.
dollar.33 The entry of technology through imports of capital goods was supposed to be
triggered by the disappearance of tariff and non-tariff barriers, and competitive pressures
were intended to drive productivity gains, with a hint of optimism about the export of
manufactures.
In order to close the external gap, capital markets were deregulated. The assumption was
that this would allow domestic interest rates to converge with international ones and thus,
via the interaction of supply and demand, the external sector would tend towards
equilibrium.
During the first half of 1995, the plan seemed successful: inflation disappeared in two
years, the Peso-Dollar convertibility survived the crises in Mexico (1995), Russia (1998),
and Asia (1997–1998), indicators showed favorable investment rates, and some sectors of
the economy reached the international technological frontier (Kosacoff, 1996). However,
given the “cheap dollar”34 and the relatively high costs of capital and labor, firms shifted
32 The privatization of the pension system included the shift from a public, solidarity-based scheme between active and passive workers to one of individual capitalization under the administration of private firms. 33 The main argument in favor of convertibility was that it would generate positive expectations and attract FDI, and anchoring one of the most important prices in the economy, the US dollar, would decrease the domestic pressure on prices. 34 The price of commodities, the entry of international capitals (in the form of debt, privatizations, and FDI, and the fact that the eradication of inflation took longer than predicted led, during this same period, to the appreciation of the domestic currency.
Chapter 2: A historical analysis of the process of competence building in an unstable system
- 68 -
from production to assembly and commercial activities, which gradually impacted the labor
market and the disarticulation of value chains (Kosacoff, 1998; Altimir and Beccaria, 1999).
On the fiscal front, despite downsizing, the public deficit did not disappear and the need
for foreign loans became a structural factor. The transfer of federal obligations to the
provinces unleashed a widespread problem of deficits in provincial government budgets,
which led to the worsening of territorial inequalities with greater impact on those areas
where the productivity of the primary sector was low productivity and the secondary sector
had not been consolidated.35
Within capital markets, distinguishing productive investments from speculative ones was
almost impossible, which eroded the solvency of the system. Meanwhile, the revenues of
foreign firms were not allocated to new investments in the country but were instead sent
abroad. At the same time, the productive sector’s demand turned to import of raw
materials, capital, and now final goods, but without the expected increase in exports. Of
course, all these situations had a deep negative impact on the external accounts, which
increased the level of external debt (both public and private) and the need for higher public
surpluses.
2.2.3.3. Science and technology institutions during the Convertibility Plan
Since the second half of the 1970s, Argentina’s S&T instructions had gradually lost much
of their government support due to both the fiscal crisis and the idea that domestic science
was not necessary for fostering technological progress. Thereafter, and until the mid-1990s,
the entire system entered into a period of successive disintegration, with sharp declines in
institutions’ budgets and capabilities. Then, in the middle of the technological revolutions
of biotechnology and information and communication technologies, the S&T institutions
had to downsize and orient their research towards the sale of technological services to the
domestic sector (testing and metrology) (Bisang, 1995).
During the 1990s, a set of policies that aimed to coordinate activity and reduce budget
constraints were implemented, but with scarce results. Coordinating institutions were
established,36 the public budget for S&T activities was increased, and competitive funds
35 This was especially true among the provinces of the north of Argentina. 36 Among these, the following are worth mentioning: the Innovation Law of 1998, the Science and
Technology Council (GACTEC), the National Secretariat of Science, Technology, and Innovation (which became a ministry in 2005), and the National Agency for the Promotion of Science and Technology (ANPCyT).
Chapter 2: A historical analysis of the process of competence building in an unstable system
- 69 -
were created and extended to enable research to be performed by traditional institutions,
universities, and public-private associations.37 Despite these efforts, the lack of
coordination between the new and the old institutions led to a high dispersion of research
activity in terms of the quantity, quality, and relevance of the projects (Chudnovsky et al.,
2006). At the same time, the lack of articulation within the public sector led to duplications,
contradictory incentives, and overall efficiency losses.
The S&T policy of the 1990s also focused heavily on the promotion of technological
innovation in the private sector. In this sense, funds to finance private-public R&D
projects were created, intellectual property regulation was reformed, and lines of credit and
subsidies were established for feasibility analysis, patenting, and organizational
improvements. However, the evidence does not support the hypothesis of an increase in
private innovation activity or a more intensive use of science and technology in the
productive structure. Nor is there evidence that support that beneficiated firms would have
behaved differently without the grants (Chudnovsky et al., 2006).
To promote the training of human resources, new universities were created, with a strong
focus on the suburbs of Buenos Aires, and new scholarships and internship schemes were
implemented. However, throughout the decade, the systematic deterioration of secondary
education had a direct impact on the tertiary and university systems as well as the quality
and quantity of labor in general.
2.2.3.4. The productive structure: specialization and static comparative advantages
The neoliberal policies of the 1990s involved a return to natural static comparative
advantages. The result was the re-primarization of the economy, the dismantling of value
chains, and the downsizing of the manufacturing industry. In agriculture, the green
revolution and the increase in the international demand for grains (due to the entry of
China into the world market but also to the expansion of the biofuel industry) led to the
specialization in the production of soybeans, extending the agricultural frontier and
replacing crop production. This impacted the livestock and other domestic consumption
goods, especially wheat and corn. At the same time, since the increase in soybean planting
and exporting did not take place in articulation with local industry, the increase in the
demand of this sector was satisfied with imported seeds, pesticides, technological packages,
37 “Competitive funding” was supposed to create a quasi-market mechanism where applicants had to compete for the supply of funds and this would improve the efficiency and effectiveness of research projects.
Chapter 2: A historical analysis of the process of competence building in an unstable system
- 70 -
and machinery, which reinforced the domestic industry’s shift from production to
commercialization (Sztulwark, 2005).
Manufacturing was the most affected sector. Large privatized public enterprises (electrical
power, gas, petroleum, communications, airlines, and water) made large initial investments
which improved the general infrastructure of the country, but in a context of massive
expulsion of labor, the disappearance of their R&D labs, and the disarticulation with local
suppliers. The rest of the firms chose what Kosacoff (1998) called “offensive and defensive
strategies.” One group of firms followed a “defensive” strategy which basically consisted of
the rationalization of the production function (expulsion of labor), the reduction of the mix
of production, and the resale of imported final goods (around 25,000 firms, 60% of the
industrial product). Conversely, a smaller group of firms followed an “offensive” strategy
with high investments in innovation activities to close the international technological gap
(around 400 companies, 40% of the industrial output).
Among the service sector, an overvalued domestic currency created incentives for the entry
of non-tradable service providers, especially in the entertainment sector.
Telecommunications also expanded following the privatization of public enterprises,
allowing the extension of telephone networks and Internet. Financial services and the
privatization of pension funds drove the expansion of retail banking, which allowed an
increase in consumer credit. A few years later, lack of scale, speculative rent-seeking
behavior, mismatches between loans in U.S. dollars and savings in Argentine pesos, and the
public deficit would make banks the epicenter of the crisis.
2.2.3.6. The labor market and domestic demand
The labor reforms of the 1990s, the labor expulsion of privatized companies, the decline in
industry, and regressive income distribution were causes and consequences of the labor
market situation during the late 1990s. During the first half of the decade, productivity
growth was based on the increase in the capital-labor relationship. By the middle of the
decade, the manufacturing sector became a net ejector of labor and the service sector could
not absorb the extra supply. The statistics speak for themselves: while in the mid-1980s
President Alfonsín faced a general strike because unemployment had climbed up to 6%, in
1998, 18% was a “normal” unemployment rate (Altimir and Beccaria, 1999).
In terms of stock, there was a sharp loss of skills associated with long-term unemployment.
In terms of flow, the fall in manufacturing activity and the deterioration of primary and
Chapter 2: A historical analysis of the process of competence building in an unstable system
- 71 -
middle school created incentives towards degrees leading to liberal professions and the
social sciences, increasing the number of accountants, lawyers, and economists at the
expense of engineers and scientists. As a result, while the manufacturing sector demanded
unskilled labor due to technification and the bias towards commercialization, the high
unemployment rate kept salaries so depressed that it was quite common to find over-
qualified workers (Beccaria, 2007).
The aggregate result was a widespread increase in poverty and exclusion that ultimately led
to the deterioration of health-care and education, the impact of which has not been fully
estimated yet (reappearance of eradicated diseases, child malnutrition, maternal mortality).
This accentuated the demand for government intervention in a public sector with severe
budget deficits. Of course, the result was a social instability and political crisis.
By 1998, the weaknesses of the Convertibility Plan were evident. As a result of these and
the financial crisis in Brazil—the main commercial destination of Argentina’s exports—
Argentina went to into the worst recession in its history. In January 2002, when
unemployment was affecting 25% of the active population, 50% of workers lacked social
protection, 50% of families were below the poverty line and four presidents had resigned in
just one month, the provisional administration of Eduardo Duhalde announced the
devaluation of the currency and the end of convertibility. After six months, this gave way
to an export-led period of growth (Lugones and Suárez, 2006).
2.2.4. “Argentina for everyone”: recovery, growth, and development
Since 2002, Argentina has been in a new growth phase that is characterized by a sustained
increase in the GDP and employment levels, and surpluses in the fiscal and foreign
accounts. The recovery has been driven by the increase in international demand for
primary commodities and the shock to the competitiveness of the manufacturing sector
generated by the new relative prices.
In terms of employment, the growth of the manufacturing sector and the expansion of
services has led to a massive incorporation of labor, which allowed rapid exploitation of the
productive capacity installed during the previous decade. At the same time, this capacity
allowed the incorporation of human resources with unequal skills levels, which, coupled
with the increase in the activity of labor-intensive sectors—basically construction—led to a
rapid improvement in income levels, which in turn increased domestic demand. The
popular ideology of President Néstor Kirchner and and his successor, his wife Cristina
Chapter 2: A historical analysis of the process of competence building in an unstable system
- 72 -
Fernández de Kirchner, summarized by the slogan “Argentina para todos [Argentina for
everyone],” has made social inclusion the cornerstone of growth. This explains the
implementation of policies to redistribute incomes38 and, therefore, a more than
proportional increase in consumption levels.
During this growth period, the knowledge supply was perhaps one of the sectors that most
benefited. Although severe articulation problems persist, the increase in public expenditure
and institutional reforms aimed at fostering S&T—such as the conversion of the Secretariat
of Science, Technology, and Innovation into a ministry in 2005—gave new momentum to
this activity. The increase in the public budget also included the increase of funds allocated
to the training of skilled labor and postgraduates. Scholarship schemes and spaces such as
science fairs and awards programs were set up in order to influence degree choices.
However, serious problems in primary and secondary education remain, both because of
infrastructure issues and the deterioration of teachers’ average skills (Porta and Bonvecchi,
2003; Porta, 2006).
Another unsolved problem in the new model is associated with the persistence of a
relatively high unemployment rate. This “hard core” of unemployment, mainly low-skilled
young people, will not be reduced automatically since many of them lack the basic
competencies that should be acquired at primary school (Beccaria, 2007).
At the micro level, most firms have sought to defend their market position and export
based on the competitiveness given by the new exchange rate, without major efforts to
develop a long-term strategy. Only a minority group has focused its investments on
improving competitiveness levels through product differentiation and technological
progress (Lugones et al., 2007). However, given the historical bias towards the domestic
market and the dependence on imported technology, and despite the increase in industrial
activity, foreign accounts continue to show the same trend as in previous periods, so much
so that, after more than two decades of free foreign exchange, the government has
advanced in the implementation of hard restrictions on this market.
The primary sector, meanwhile, was able to take advantage of the commodities boom,
reaching record harvest levels in a context of rising prices. However, as its production
impacts the level of wages, the expansion of the primary sector, the links of which with the
38 The main policies in this regard are the child benefit program, the pension system for housewives, the price control on basic goods, and the sustaining of transportation and public services subsidies.
Chapter 2: A historical analysis of the process of competence building in an unstable system
- 73 -
local manufacturing sector remain virtually nonexistent, has reactivated distributive
pressures and led to social confrontation.39
Within the service sector, policies promoting high-tech activities implemented in previous
periods and sustained during the current one began to mature, leading to the emergence
and growth of software, computer services, and biotechnology firms (Borello et al., 2005;
Gutman and Lavarello, 2010; Stubrin, 2011). However, the sector is not large enough to
bring about a change in the productive structure towards goods and services with higher
knowledge content. The financial system, meanwhile, turned to the offer of credit for
consumption, and banks remain far below international standards, which implies that the
manufacturing sector still faces high financial costs.
In summary, this new phase resulted in the reappearance of discussions about the
possibilities of long-term growth, debates that were accentuated in September 2008, when
the recent global financial crisis highlighted the need for a new strategy. Despite
improvements in the macro indicators (GDP, exports, employment), the specialization
pattern shows no signs of change—on the contrary, the productive structure continues to
specialize in goods with low technological content, and distributive pressures are still a
cause of social instability.
2.3. Innovation and competence building: structural characteristics
Given the history presented above, this section aims to analyze the process of competence
building in Argentina’s NIS in order to identify those characteristics that arise from
historical processes and impact the current stage of development. To do so, Figure 2.2
summarizes the main features of each of the dimensions explained in the methodological
section.
39 In 2008, the government tried to implement a self-adjusted export tax to automatically adjust the level of export taxes to international prices in order to control the domestic inflation. This measure provoked strong opposition not only in the primary sector but within an important part of society and led to several demonstrations against President Cristina Fernández de Kirchner. Although the measure was not approved by Congress, the social upheaval it caused led to a fracture in society along the “rural vs. industrial” divide, which mirrored the positions of the ruling party and the opposition.
Chapter 2: A historical analysis of the process of competence building in an unstable system
- 74 -
Figure 2.2: Summary of the history of the Argentinean innovation system
AEM (1880–1930) ISI (1930–1976) Liberalism (1976–2002) Current model (2002-
onwards)
Supply and demand
Firms
Rise of exporter agriculture
Development of transport and
communications Incipient development of manufacturing and
urban services
Rise of domestic-market oriented manufacturing Retreat of agriculture
Development of industry-oriented services
Homogenization of skills
Reprimarization based on comparative advantages Drop-in manufacturing
and privatization Development of financial
activities
Recovery of manufacturing with
increase in exports due to the exchange rate
Regional bias of exports.
Supply Chain
Integration into foreign value chains
Importation of technology into
agriculture Increase in the labor
supply
Lack of articulation between agriculture and
manufacturing Importation of
technology Incipient articulation with local
value chains Full employment
Disarticulation of local value chains
Articulation with foreign value chains in low added-
value segments Loss of labor skills
High unemployment rate
Increase in production due to domestic growth.
Persistence of disarticulation between the productive sectors
Low unemployment rate
Demand
Increase in domestic demand due to
immigration and improvements in
income levels.
High domestic demand satisfied by domestic
production Low sophistication of tastes and preferences.
High domestic demand satisfied by imports
Concentration of incomes. Drop in income levels,
demand focused on basic needs
High domestic demand partially satisfied by local production, increase in
pressure on imports
Science, technology &
education infrastructure
S&T system Lineal model of
innovation Creation of the main S&T institutions
Budget restrictions Competitive funds
Loss of qualified human resources
Disarticulation within the S&T system
Competitive funds and budget increase
Improvements in the articulation with firms
persistence of disarticulation within the
S&T system
Education
Creation of the national public
university (liberal degrees)
Universalization of primary education
Creation of technological degrees
Low articulation with the productive sector
Universalization of secondary education Consolidation of the public universities
Predominance of social sciences, research, and
teaching with low articulation with the
productive sector Deterioration of primary and secondary education
Predominance of social sciences and liberal
degrees Disarticulation with the
demand of the productive sector
Public Sector
Liberalism and promotion of immigration
Investments in communication and
transport
Welfare state Industrialization
Social rights and inclusion
Neoliberalism Deregulation and
downsizing Market-oriented reforms
Industrial policy Promotion of exports
Intervention in the markets
Income redistribution
Rest of the world
Export of NR-based commodities
Importation of technology
Importation of technology
Export of NR-based commodities
Importation of technology and final goods
Increase in NR-based and manufacturing
exports International crisis
Incentives to interactions
Free trade Market relationships
Strong interaction with foreign markets (UK
and US)
Protectionism Relationships ruled by
access to public promotion programs
Labor unions Predominance of
domestic interactions
Free trade Market-based relationships
Prices ruled by the evolution of the US dollar Mismatches between local demand and production
Protectionism: exchange rate and importation
restrictions Interactions with the
knowledge infrastructure based on the demand of
skilled labor.
Chapter 2: A historical analysis of the process of competence building in an unstable system
- 75 -
In stylized terms, the main characteristics of Argentina’s NIS are the same as those
predicted by the literature (Chapter 1): a specialization pattern based on static comparative
advantages, a history of economic, social, and political instability, and a weak articulation
between the productive sector and science, technology, and educational institutions.
However, history has also allowed some areas to emerge where technological development
is at the core of private strategies and where the knowledge creation process leads to
innovation by means of an articulated net of organizations. These virtuous cases cannot be
appreciated when looking at aggregate levels but they certainly appear when micro-
heterogeneity is accepted. The structural features which explain the basis of the system
nowadays—included the “exceptions” that account for the need for an alternative
interpretation—are discussed in the following sections.
2.3.1. Industrialization versus comparative advantages
The historical overview presented in Figure 2.2 shows that the dynamic of competence
building has been based, since 1880, on the incorporation of imported technology with
weak linkages with the local S&T system or the productive structure. Most of the
explanation can be found in the alternation between the promotion of free trade (the
primary sector) and protectionism (the manufacturing sector), without an explicit policy of
coordination between them, such that the growth in one sector could drive the
development of the other. Every time public policy chose to foster one side, the
technological gap was so high and the animal spirits “were so sure” that the policy would
not last long that the catching-up had to be done by importing capital goods, and
investment decisions had to be based on short-term benefits.
In 1930, the primary sector could have closed the technological gap and moved forward on
intensive exploitation of the land by means of the local production of capital goods but
public policy was centered on the need to develop the manufacturing sector by means of
“easy” industrialization and the import of capital goods. In 1976, the opening and
deregulation of the economy and the fiscal problems worked against those manufacturing
sectors that had accumulated skills and seemed to be moving towards the endogenous
creation of knowledge (aviation, automotive, agricultural machinery, basic genetic
manipulation, pharmaceutical, and chemical industries) (Katz and Kosacoff, 1998). During
the 1990s the primary sector reached the international border again, which allowed an
increase in extensive and intensive crops. However, it was not able to bring about an
equivalent effect in the manufacturing sector. In contrast, the incorporation of technology
Chapter 2: A historical analysis of the process of competence building in an unstable system
- 76 -
came through importing, so even directly related sectors (machinery and basic inputs)
could not take advantage of the growth boom.
The rest of the industry also failed to develop upstream linkages. The deepening of the
economic opening of the 1990s and the increase in competitive pressure led to the quest
for short-term technological improvements by means of importing machinery and
equipment, almost neglecting the requirements of local adaptation, especially to the
domestic scale. This was a strong disincentive for local capital industry, which gradually
became an importer and reseller of technology (Katz and Bernat, 2011). Subsequently, the
cheapening of imported goods led to the productive mix being complemented by imported
final goods, so that the only value added was associated with marketing and, at best, local
adaptations.
Nowadays, the overall result of this dynamic is a productive structure that can be
characterized more as a sum of isolated sectors than as an innovation system. At the same
time, given that low technological complexity characterizes the country’s productive
structure and the fact that the introduction of technology is based on the importation of
machinery and equipment (with some minor exceptions that fail to alter the selection
environment), labor demand is biased towards low-skilled employment which feeds back
into the low capabilities of the productive structure. Within this context, the multiplying
effect of public expenditure on knowledge creation is reduced, and qualified human
resources—trained using public funds—are largely absorbed by a disarticulated S&T
system—also public—which in turn reinforces the endogamic logic of the knowledge
infrastructure.
2.3.2. Growth versus recession
Economic cycles have also left their structural imprint on Argentina. In theoretical terms,
contexts of uncertainty and high vulnerability impact the decision-making processes about
investments. At the same time, highly uncertain macroeconomic contexts reduce the cost
of delaying investments versus the expected benefits. As a consequence, strategic decisions
and long-term technology development projects are postponed. Given the intrinsic
uncertainty of innovation processes, the risk perceived by the firms may outweigh the
expected benefits (Porta and Bonvecchi, 2003; Kosacoff and Ramos, 2006). In contexts
like these, radical innovations are seldom a rational choice.
Chapter 2: A historical analysis of the process of competence building in an unstable system
- 77 -
In terms of innovation, unstable environments generate incentives for short-term
improvements based on the acquisition of capital goods (the certainty of machinery versus
the uncertainty of R&D), organizational innovations aimed at rationalizing the labor-
capital relationship, cost reduction (price competitiveness), and the overexploitation of the
installed capacity (instead of the quest for expansion) (Frenkel and Rapetti, 2011).
The evolution of the exchange rate and the price index are two other factors that account
for macro instability and impact how agents perceive innovation. Between 1980 and 2002,
Argentinean entrepreneurs had to face an hyperinflation peak of 5,000% in 1989, a 300%
devaluation of the exchange rate within only two months in 2002, and an asymmetric
“pesification” where debts were converted into the domestic currency while deposits were
first confiscated and repaid in dollars almost a decade later.
From the private perspective, uncertainty about the evolution of inflation and the price of
the U.S. dollar distorted domestic prices. The “expensive dollar” during the ISI years was
perceived as a protectionist measure against imports rather than an incentive to export,
and the “cheap dollar” during the Convertibility Plan was perceived as a way to maximize
incomes by locally manufacturing simple technological goods and importing more
complex ones.
The lack of monetary stability and the fragility of the financial system contributed to more
value being assigned to accounting and financial skills than to production and organization.
During ISI, firms’ profitability was associated with obtaining promotional benefits,
negotiating tariff protection, and creating profits from the speculation around the U.S.
dollar and interest rates (“textile and financial” was a common economic activity during
those days). During the Convertibility period, the complex tax structure (a result of the
public deficit) led to real benefits being associated with how incomes were reported, and to
financial rent being more secure and profitable than productive (since the capital market
was completely deregulated). Meanwhile, besides the loss of skills associated with
productive disintegration, large companies, privatized enterprises, and the local subsidiaries
of multinational firms became importers of technology, local resellers, or assemblers. This
was the case for automobiles, telecommunications, electrical power, and, especially,
suppliers of the primary sector.
As a result, Argentina’s NIS is highly prepared to respond to abrupt changes in macro
dynamics but it is unable to see beyond the short term. The most important intangible
Chapter 2: A historical analysis of the process of competence building in an unstable system
- 78 -
institutions are those associated with taxes and finances, and innovations in those activities
become key to the firm’s survival: how to determine final price, make profits from the
interest and exchange rates, sustain purchase power by avoiding the payment of taxes, etc.
Changes in these institutions were so frequent that they demanded constant efforts to keep
up with the regulation. The result was less available resources to innovate.
2.3.3. Public versus private goods
Besides its role as a promoter of industrial development, the Argentinean government has
had a leading role in the development of the science, technology, and educational systems.
Of the total expenditure on S&T in 2009, the Argentinean public sector accounted for
67% of investments, with an increasing trend over the past decades (in 1994, this ratio was
45%). This level of spending contrasts with the values recorded in regional peers like Brazil
(54%) (RICyT, 2012). Additionally, 55% of the university system is accounted for by
public universities, which in turn account for 81% of undergraduates and graduate
students (SPU, 2010). This relationship directly impacts both the stability of the system
and the direction of knowledge creation.
Firstly, expenditure levels are determined by the evolution of public accounts. Given the
instability that has characterized the country, S&T projects which require a long-term
horizons (and funding) are almost impossible to implement. At the same time, the deep
fiscal crisis led to recurrent drops in expenditure levels, not only because of the lack of
actual funds but also because of the reallocation of resources towards more urgent
problems. This dynamic of budget expansion and contraction impacts skills and
infrastructure (obsolescence of equipment, brain drain, closing of public S&T institutions,
etc.).
Although, theoretically speaking, the lack of funding is a powerful incentive to interact
with the private sector (to access funds), in a context where private innovative activity is
low, the technological complexity of the linkages will be low too. That is, incentives to link
are generated, but the complexity of knowledge that flows within the interaction is reduced
and is more associated with routine activities than with the quest for new products and
processes (Lugones and Suárez, 2006; Suárez, 2007). The local outcome of this situation
was the emergence and continuing of a double dynamic within the S&T system:
institutions sold low-tech services connected to the manufacturing sector in order to get
funds to support high-tech research, far from manufacturing requirements. Meanwhile, the
Chapter 2: A historical analysis of the process of competence building in an unstable system
- 79 -
manufacturing sector chose to pay cheap local S&T services and import the more complex
technologies.
A third component of the disarticulation of the system is the lack of coordination between
different public policies. During ISI, pharmaceutical and chemical research was promoted
but industrial policies focused almost exclusively on the textile and metalworking
industries. During the 1970s, S&T policy was directed towards nuclear research but
hydraulic power plants were developed with public funds. During the 1990s and in the
middle of the boom of the agricultural sector, the S&T institution most closely connected
to agriculture, INTA, faced severe budgetary constrictions. Nowadays, there has been an
increase in the number of scholarships for doctoral training but neither the S&T system
nor the private sector is able to absorb the amount of human resources that are being
trained. At the same time, given this weak interaction between S&T institutions and firms,
financing such institutions cannot be use as a channel to foster private investment.
In relation to the education system, the lack of coordination between degree programs and
labor demand, and the prominence of the liberal degrees and social sciences over
engineering and natural sciences are a historical feature. During the ISI, the “petty
bourgeois” ideology projected social mobility based on the liberal professions. During the
1970s, the fall of industry also generated incentives to following liberal careers, where the
possibility of self-employment was an insurance against the lack of labor demand. This
idiosyncrasy and the ability to work independently were accentuated during the 1990s,
given the disincentives emanating from the collapse of industrial production. At the
beginning of the 21st century, the deterioration of middle school education created further
cognitive distance between future skilled employees and the sciences. As a result, 44% of
graduates from public universities had studied some social sciences, while medicine, law,
architecture, economics, and administration accounted for 62% of graduates’s chosen
degree courses in 2008 (SPU, 2010).
The weak link between the S&T system (including universities) and the productive
structure is also explained by the level of endogenous competencies. Throughout
Argentina’s recent history, firms’ endogenous competencies were associated with the
incorporation of external technology, which means low technological complexity of the
activities carried out internally. The S&T system, marked by the pursuit of public goods
and articulated with the international community, stands at more complex levels in terms
of technology, with no incentives to apply knowledge to problem-solving (less complex)
Chapter 2: A historical analysis of the process of competence building in an unstable system
- 80 -
activities. In simpler terms, the S&T system has no incentives to search for solutions for
simple problems, and the industry is incapable of transforming the created knowledge into
innovation since it is too complex.
Despite this general situation, there are some successful cases which confirm that, in those
areas where interactions do exist, the impact has been positive for both firms and S&T
institutions (Suárez, 2009; Yoguel and Suárez, 2011). These interactions are explained by
the level of firms’ endogenous capabilities, the availability of public and private resources,
and parties willingness to interact. They also account for the existence of different
situations and the fact that even when facing the same incentives, some organizations
behave differently.
2.3.4. Virtuous behavior versus rent-seeking behavior
The historical analysis of Argentina’s NIS also explains some features that have
characterized the skills accumulation processes, which has been repeatedly identified by
several authors over the last 15 years (Anlló et al., 2007; Erbes et al., 2004; Milesi, 2006;
Suárez, 2006; López and Arza, 2008; Lugones et al., 2008). These are: i) low commitment
to innovation; ii) innovation investments biased towards the acquisition of embodied
technology (over endogenous forms of knowledge creation); iii). a high level of firms with
innovation results but a reduced scope of the innovations; and iv) high level of interaction
with the innovation system but low technological complexity of linkages.
Thus, despite the changes in the national and international scenes within the current
economic model, firms seem to reproduce the innovation patterns of earlier periods.
During ISI, competition was based on access to promotional benefits; during the
Convertibility Plan, it was based on importing final goods; in the current model, it seems to
be based on the exploitation of the cost advantage given by the devaluated currency and
the low external competition given the explicit protectionist policies.
However, like in other periods of growth, there are firms that show different innovation
results, despite having been exposed to the same incentives and the same history. In this
sense, there is a fifth feature of the production structure, which has been less highlighted
by the literature: the inter- and intra-sectorial heterogeneity, not in the sense of variety but
in terms of different productivity levels and virtuous strategies. During ISI, there were
firms which decided to seek genuine productivity gains and export, despite the strong
incentives given by a protected domestic market. These firms were then better prepared to
Chapter 2: A historical analysis of the process of competence building in an unstable system
- 81 -
deal with the process of opening-up during the following period and lessen the impact of
the domestic cycles. Of course, the bulk of the productive structure opted for the
exploitation of the domestic market on the basis of foreign protection (high prices and low
productivity) and they were in a weak position during the opening-up process (and even
disappeared from the market). During the 1990s, some firms improved their productivity
via the incorporation of external technology but combined it with endogenous knowledge
creation. These firms were then in a better position to deal with the 1998–2001 crisis and
to return to the path of growth once the recession had been overcome. In contrast, those
firms that followed a strategy based on the resale of imported goods almost disappeared
from the market after the devaluation of the currency. During both ISI and the 1990s, the
first type of firm accumulated skills and advanced along technological learning paths, while
the second type was on the verge of disappearance since the accumulated skills were
useless for the new environment.
From a more historical perspective, this heterogeneity is consistent with that observed by
Katz in 1976, Kosacoff in 1998, and Katz and Stumpo in 2001. Katz (1976) also warned of
the existence of firms where the level of investment in R&D was proof of innovative
behavior based on the endogenous generation of knowledge. Kosacoff’s (1998) proved the
existence of differential responses across firms’ sectors and sizes, as a result of which some
firms closed the technological gap from the intensive creation of knowledge. In 2001, Katz
and Stumpo (2001) noted that, for some sectors, the technological gap was narrowed and
was explained not only by the existence of large firms subject to a long history of
protection and promotion, but also by the mixed investments in embodied and
disembodied technology. More recently, similar findings about heterogeneity are contained
in Lugones et al. (2008) and Milesi (2006) for the manufacturing sector and in Borello et al.
(2005) and Roitter et al. (2011) for services. In all cases, the authors highlight the
heterogeneous set of responses to the micro and macro incentives and the importance of
complementing the macro analysis with micro-studies of innovation.
2.4. Preliminary conclusions
Throughout this chapter, the Argentina’s recent history in terms of the dynamics of
knowledge production and appropriation has been analyzed. The main objective was to
identify what Katz and Kosacoff (1998) have called “local idiosyncrasies,” which are those
patterns of behavior that are repeated over time across the entire productive structure. In
this sense, a history of macro instability, the oscillations between industrial protectionism
Chapter 2: A historical analysis of the process of competence building in an unstable system
- 82 -
and liberalization, and the disarticulation of the S&T system have led to an NIS which, as
Sutz and Arocena (1999) argue, is far from presenting a systemic dynamic.
Another element which stands out from the analysis is the fact that some of Argentina’s
underdevelopment characteristics do not seem to be part of a process towards a more
developed situation. Given the history described above, it is hard to believe that
organizations will behave as if instability, changes in the rules of the game, and recurrent
cycles of growth and recession were going to disappear. On the contrary, organizations
have learnt how to deal with this environment and they behave accordingly.
At the firm level, the presented analysis has shown how a particular history led to a specific
set of features that determines firms’ average response to environmental incentives and
public policies, which are the same average responses Dutrenit and Katz (2005) observed
for Latin America as a whole. However, it also showed that there are different types of
innovative behaviors, some of which deal better with the described environment. This
micro-heterogeneity has generated niches of successful innovative dynamics, which
represent an alternative for long-term development.
When one looks for explanations for these successful behaviors, some positive features of
the Argentinean system stand out: i) the existence of some educational institutions with a
long, high-quality trajectory ; ii) a universal, free, and public education system and a critical
mass of qualified human resources; iii) the existence of some S&T institutions that have
expertise in delivering services to the private sector; and iv) new competitive firms
emerging and competing in the most dynamic markets. All these positive characteristics of
Argentina’s NIS have received less attention in the literature, and have mostly been
relegated to case studies or just described as “exceptions” to the general trend.
In this context, and following the NIS approach, those analyses based on the concept of
catching up fail to understand the complex nature of competence building as well as
structural characteristics. In the case of Argentina, this means the unstable macroeconomic
environment. Conversely, those studies based on structural features fail to identify micro-
heterogeneity and its potential. If some firms learnt how to persistently innovate in this
type of environment, then identifying and characterizing them could shed light on how to
foster and multiply them, accepting the complexity of the system, its structural
characteristics, and its recurrent instability. Understanding how firms behave and why they
react differently is at the core of the thesis and the objective of the following sections.
Chapter 2: A historical analysis of the process of competence building in an unstable system
- 83 -
SECTION II - The innovative strategies
Introduction to Section II
According to what was discussed in the previous section, the innovation process at the
micro level is the sequence of decisions and actions that leads the firm to transform
knowledge into innovations. The possibility of an innovation emerging, even when it is a
chance event, depends on the firm’s learning capability, which is subject to the particular
national innovation system (NIS) where it operates.
Within the NIS approach, there are robust explanations about the firms’ innovative
dynamic in Lundvall’s, Nelson’s, and Freeman’s work, as well as a set of other theoretical
approximations that more or less coincide with their approach. Besides criticizing the
unrealistic neoclassical assumption about the representative firm, these studies
acknowledge the complexity of the firm’s innovative behavior as well as the heterogeneity
of any productive structure. However, as was also discussed in the previous section, this
heterogeneity is associated more with the variety of the system than with the existence of
different competitive strategies, where innovation is just one of many possible paths for
facing competition.
Within the context of unstable environments, micro-heterogeneity is even greater since
multiple ways of competing, multiple levels of productivity, the weak institutional set-up,
and changes in the rules of the game exponentially increase the heterogeneity of
microeconomic answers. As such, the interaction between path dependence and possible
futures becomes key to understanding firm performance. In this respect, the
abovementioned contributions to the NIS approach agree on the importance of the
environment and path dependence processes but they fall short when explaining how the
innovative dynamic at the firm level leads to the firm’s specific path creation. In other
words, this literature sustains that the innovation process triggers feedback and
accumulation processes that increase the firm’s capabilities, but it does not explain how
these processes could actually alter the firm’s evolution.
Introduction to Section II
- 84 -
I will argue that path dependence and path creation are a matter of strategic decision at the
firm level, and that they determine the observable behavior of the firm. Firms take
decisions regarding how to face the selection process based on their environment,
capabilities, and resources. These decisions can be interpreted ex-post as the firm’s strategy,
and innovation is just one possible way of facing the process. Feedbacks from strategy
implementation and environmental evolution will impact the set of decisions that make up
the strategy, forcing the firm to adjust it. In this sense, innovations are the result of a
dynamic process determined by several firm-level decisions which are both cause and
consequence of the capability of learning from past and present actions as well as from
signals from the environment.
If the environment changes, the firm will rethink its strategy and take new decisions
regarding how to seek out competitive advantage, including again whether to follow an
innovative strategy or not. The higher the level of the firm’s capabilities, the greater the
possibility of taking advantage of (or avoiding) the impact of environmental change. As we
shall see, this explains why “good” firms can be found in “bad” environments and to what
extent firms are able to reverse the lock-in effect of past competitive trajectories.
Therefore, the objective of this section is to discuss the state of the art in the matter of
innovative firm behavior, its determinants, and dynamic impacts in order to provide a
theoretical approximation to the innovative dynamic of the firm that is capable of
accounting for path dependence and path creation, and which ultimately explains the
existence of different types of micro-heterogeneity. In doing so, this section aims to
contribute by zooming in on specific aspects of the NIS approach by providing theoretical
elements to understand the relationship between the NIS and micro-heterogeneity. To do
so, the strategy-based approach will be presented, discussed, and linked to firm
performance in Chapter 3. In Chapter 4 a conceptual approximation to innovative
strategies will be proposed and the hypothesis to be tested will be discussed. This will
establish the conceptual basis for the methodological approach and the testing of the
hypothesis, which is the subject of Section III.
- 85 -
Chapter 3: The strategy-based approach to
firm behavior
Introduction
From an evolutionary perspective, there is general consensus about the fact that firms have
specificities that make them unique and that, because of this, the productive structure will
be made up of a heterogeneous set of agents (Penrose, 1959; Nelson and Winter, 1982;
Lundvall, 1992; Freeman and Soete, 1997; among others). This literature also shares an
acknowledgement of the learning processes as a key element to explaining how firms
transform knowledge into goods and services that allow them to survive and succeed
within the market. According to this view, innovation is a dynamic process in which path
dependence, feedbacks, and competence building explain firms’ performances.
Several empirical analyses have tested the theoretical postulates of this literature (Arundel
et al.., 2007; Leiponen and Drejer, 2007; Clausen et al., 2011; Srholec and Verspagen, 2012;
Raymond et. al, 2010; among others). These studies provide evidence about the impact of
learning and capabilities and how the firm faces the environment. Another characteristic of
this literature is the static analysis of the firms’ behavior. These studies can explain why a
firm performs better but not how it reacts to this increase in income or profits. Likewise,
they can explain how a firm’s innovative strategy is based on high capability levels, but not
how the learning processes that arise from these capabilities feed back into the innovative
strategy. In other words, this literature can explain the impact of path dependence but not
why and how firms can avoid it. This lack of explanation is the driving force behind this
chapter.
The objective of this chapter is to shed light on the fact that sometimes firms decide not to
innovate and survive the market all the same, and to present certain features to explain why
the concept of firm strategy is key to understanding micro-heterogeneity and how
successful firms can be found in unstable environments. The final goal is to propose a
theoretical approach to the firm capable of accounting for the path dependence (David,
2007) and the path creation (Garud, et al., 2010), and the learning processes that take place
between them. To answer this question, in Section 3.1 the main evolutionary approaches to
the firm behavior are presented. In Section 3.2, the key contributions and limitations of the
approaches to the understanding of innovation are analyzed. In Section 3.3, the strategy
Chapter 3: The strategy-based approach to firm behavior
- 86 -
-based approach is proposed, in order to make a contribution to the dynamic
understanding of the firm. Preliminary conclusions are provided in Section 3.4.
3.1. Capabilities, resources and firm behavior
3.1.1. Evolutionary approaches to firm behavior
To paraphrase Penrose (1959), a firm is not an objective entity that can be defined without
a context. A firm is a legal entity, a group of buildings, an aggregation of financial
operations, an economic agent, etc. The definition of a firm requires a theoretical
framework and a research objective. For the purpose of this study, and from an
evolutionary perspective, firms are one type of organization which interacts with other
organizations and exists within a particular environment.
Given this definition, there is wide consensus regarding the profit-seeking nature of the
firm, with “profit” being used in a broad sense to include, for instance, the quest for
survival, growth, or diversification (Penrose, 1959; Nelson, 1991a). In their quest for profit,
firms are organizations that produce and compete in the market, where their competitive
position will determine the actual level of rent (Nelson and Winter, 1982). Therefore, firm
behavior is associated with decisions regarding how it seeks to profit from the competition
process in the market.
In dynamic terms, past decisions will affect present ones since firms are subject to path
dependence. According to David (2007), path dependence processes are those with
trajectories shaped by their own history. This means that the process has intrinsic
mechanisms that reinforce its trajectory (feedbacks) with increasing returns (accumulation).
At the firm level, feedbacks and accumulation processes lead both to the emergence of
learning effects, which improve the process itself, but also to lock-in effects, which work
against the possibility of finding new profitable trajectories. However, at some point, and
given specific conditions—Davis’ (2007) contingencies; Schumpeter’s (1947) creative
reactions—some actors can set a new trajectory in motion, usually referred to as path
creation (Lovio et al., 2011).
Path creation is the result of the fact that individuals and organizations are not passive to
their path dependence. They try to foresee the future, to anticipate where their path is
taking them, and to know which actions should be taken in order to reach a desirable
outcome (Garud et al., 2010). In this sense, path creation is an intrinsic element of the firm
Chapter 3: The strategy-based approach to firm behavior
- 87 -
which is a function of the organization’s ability to master its own trajectory. It is an
endogenous force within the firm which depends on the latter’s capabilities for searching
for new profitable options (which are different from Davis’ exogenous contingencies that
trigger the firm reaction). It is not a deviation from the existing path (Schumpeter’s creative
reactions) but a process that leads to the creation of a new path.
Based on this basic understanding of the firm, three complementary explanations about
firm behavior can illuminate the decision-making process that connects path dependence
with path creation: the capability-based approach, the resource-based approach, and the
knowledge-based approach.40 The capability-based approach groups all explanations based
on Nelson and Winter’s (1982) concept of routines and capabilities (Cohen and Levinthal,
1990; Nelson, 1991b; Teece and Pisano, 1994; Langlois and Robertson, 1995; Narula, 2003;
Antonelli, 2008). The resource-based approach also focuses heavily on capabilities, but on
the basis of Penrose’s (1959) idea of the firm as a set of resources and assets (Srholec and
Verspagen, 2012; Wernerfelt, 1984). The knowledge-based approach is closely related to
the fundamentals of the NIS approach and explains firm dynamics as the result of an
interactive process of knowledge creation and accumulation, subject to external
environmental factors and internal resource endowment, in the same sense as the resource-
based approach (Arundel et al., 2007; Lundvall and Lam, 2007, Metcalfe, 2002).
3.1.1.1. The capability-based approach
According to the capability-based approach, firms are defined as a repertoire of routines
and capabilities the main objective of which is to exploit knowledge in order to make
profits. Combining the concept of routines with the concept of capabilities, this approach
sustains that the firm’s dynamic is the result of the interaction between two different
processes, one endogenous to the firm—the adaptation and creation of routines—and one
exogenous to it—the selection environment. The following paragraphs examine these
processes in detail.
The selection environment is an exogenous force affecting the type of decisions firms can
make. According to Nelson and Winter (1982), Schumpeterian competition can be
40 Given its roots in the evolutionary contributions of Nelson and Winter, some authors refer to the capability-based approach as the “evolutionary ideas of the firm” and differentiate it from the behavioral approach——here presented as the resource-based approach (Leiponen and Drejer, 2007; Rahmeyer, 2007; Teece and Augier, 2007; Srholec and Verspagen, 2012). However, both the resource-based and knowledge-based approaches can also be referred as evolutionary given the evolving and complex nature they assign to the firm, as well as their interpretation of capitalist competition and technological change. This is why different terms are being used to refer to these approaches in this dissertation.
Chapter 3: The strategy-based approach to firm behavior
- 88 -
understood as a process of environmental selection in which firm behavior and market
competition are jointly determined and they both lead to the selection of price (and
productivity) levels and firm investment decisions. A firm’s characteristics (assets, financial
situation, capabilities, technology) determine its productivity levels. The aggregate of firms,
consumers, and suppliers determines market characteristics and, especially, price levels.41
Consequently, taking into account the firm’s productivity and the market conditions
(prices), the firm takes decisions about its own market price, and hence about its
profitability levels. Within the following period, profitability levels will lead it to adjust
investment decisions regarding its own expansion or contraction, which will ultimately lead
to new levels of productivity, and the process will start all over again. In this way, the firm
will seek productivity improvements in order to stand out from the competition, and the
decisions it makes will impact that competition’s characteristics and the maximization of
profits. This is what Nelson and Winter (1982) call the “simultaneous and evolving process
of market selection,” which they summarized as the “selection process.”42
The endogenous process arises from the interaction between the firm’s routines and
capabilities. As was presented in Chapter 1, routines are standardized forms of behavior
that determine the productive, financial, and innovative dynamic of the firm. They reduce
the cost of deliberation, communication and reaction to the regular dynamic of the firm
(Nelson and Winter, 1982). The level of capabilities arises from the formal education of the
individuals together with their personal and collective trajectory (Cohen and Levinthal,
1990; Narula, 2003; Teece and Augier, 2007). The aggregate of routines defines what the
firm is actually doing, and the level of capabilities what the firm could possibly do (Langlois
and Robertson, 1995).
The level of capabilities will determine the information and knowledge the firm can absorb
from the environment and how and in what direction routines could be improved (Cohen
and Levinthal, 1990; Narula, 2003). The quest for improvements to routines—either
deliberately or as a by-product of problem-solving—is called the innovative process and it
will determine how the firm will face the selection environment in the next period (Nelson
and Winter, 1982; Rahmeyer, 2007; Teece and Augier, 2007).
41 We will see later in this chapter that this selection environment is also made up of other organizations affecting the market competition. 42 Also following Nelson and Winter (1982), the firm decision process has random elements so the firm cannot know the result until it has faced the selection process. This means that investment decisions (innovation decisions included) are part of the cost function of the firm and affect its profit levels. Therefore, innovation projects take part of the selection process in the form of increased production costs until they become innovations and allow the firm to reach extraordinary profits.
Chapter 3: The strategy-based approach to firm behavior
- 89 -
A key element of this approach is the concept of dynamic capabilities (Nelson and Winter,
1977; Nelson, 1991a, 1991b; Teece and Pisano, 1994). In theoretical terms, dynamic
capabilities are a firm’s ability to master assets and ordinary capabilities (absorption,
connectivity, technological, and commercial capabilities) in order to cope with the selection
environment. Since the environment is not stable, the firm has to constantly adapt its
competitive dynamic to these changes by reconfiguring its existing capabilities. Therefore,
the firm’s dynamic capabilities will determine how fast, accurately, and successfully it
acknowledges and reacts to these changes. In other words, ordinary capabilities lead to
knowledge being transformed into innovations, while dynamic capabilities lead this process
happening persistently (Teece and Augier, 2007).
Nelson (1991b) provides a more practical explanation of dynamic capabilities. Much more
influenced by management literature, and searching for an explanation of firm-level
heterogeneity, the author argues that dynamic capabilities allow the firm to cope with the
uncertainty that exists before and during the innovation process. Dynamic capabilities can
be defined as the co-evolution of the firm’s strategy, structure, and core capabilities. The
strategy is the set of stable heuristics disseminated throughout the organization from which
innovation processes are selected. It is the set of choices regarding which objectives the
firm will pursue and how it intends to accomplish them. The structure is the specification
of the strategy and includes the assets and how they are organized. While the strategy is a
set of broad commitments, the structure defines how the firm will be organized and
governed in order to achieve the selected commitments (Chandler, 1966). Finally, the core
capabilities determine the range of possible (profitable) choices the firm can make but also
the actual results that will arise from the exploitation of the assets. If the strategy changes,
the structure will also change and capabilities must be adapted. At the same time, since the
firm’s competition and productive dynamic will trigger feedback processes that will
enhance capabilities, the range of strategic options will also increase, allowing new
adjustments or changes to the current strategy.
Two conclusions from this approach shed light on the relationship between the
environment and micro-heterogeneity. The first has to do with the apparent contradiction
between routines and adaptation. In order to cope with uncertainty and organize the
productive process, the firm has to standardize procedures and practices, that is, to
establish routines. However, routines have to be counter-balanced with learning processes
which improve capabilities and allow for changes otherwise, organizational inertia could
Chapter 3: The strategy-based approach to firm behavior
- 90 -
work against adaptation and, of course, survival (Langlois and Robertson, 1995). As a
result, while routines are associated with path dependence, learning is associated with
adaptation and hence path creation. Consequently, besides the importance of routines to
comprehending the firm’s dynamic, the creation and accumulation of capabilities are key to
understanding the firm’s unique behavior and, therefore, micro-heterogeneity.
Secondly, and following Nelson’s (1991b) conclusions, given the number of variables that
affect the “ever-evolving, constant, simultaneous processes of selection and adaptation”, a
firm cannot simply “select the best strategy”. It is not a linear but an interactive process in
which feedbacks from the environment, the performance of the structure, and the
capabilities lead to constant adjustments and improvements in the firm strategy (and vice
versa). The way the strategy, structure, and capabilities will evolve can hardly be foreseen.
On the contrary, the observable outcome of this interaction is the result of process of
learning and adjusting. The evolving and complex nature of the interaction between
strategy, structure, and capabilities explains why there are a multiplicity of heterogeneous
situations at the firm level, which also change over time. Analysts can characterize the
observable outcome—the firm’s innovative structure and its performance in the market—
but they cannot observe underlying strategy.
3.1.1.2. The resource-based approach
The resource-based approach—also referred to as the behavioral approach—groups
together all analyses that define the firm as a pool of assets which are exploited in order to
attain profits. This approach is based on Penrose’s (1959) definition of the firm as a
collection of productive resources that are allocated to different uses and will perform
differently according to how human resources use them. For this author, as the
environment changes, the firm’s decisions cannot remain the same, but they must always
allow its resource allocation to adapt to cope with these changes. At the same time, since
the production and commercialization processes generate new resources, new decisions are
required. In the same sense as Nelson, Penrose argues that interactions with the
environment and within the firm trigger learning processes, which increase the firm’s asset
and knowledge levels and improve its possibilities of long-run profits.
At the micro level, the firm’s dynamic is the result of the interaction between market
strategy and internal organization, its capabilities, and its resources (Penrose, 1959). Once
the firm’s market strategy and internal organization have been established, the firm builds
Chapter 3: The strategy-based approach to firm behavior
- 91 -
capabilities and competences—strategy, structure, and core capabilities in Nelson’s terms—
and selects resources to deal with the competition process. The dynamic of resource
exploitation—how new resources are acquired—and the impact of market selection will
determine the firm’s competitive strength.
Within this framework, the firm’s innovative dynamic arises from two key aspects of its
resources: their indivisibility and their differential performance. On the one hand, as long
as assets and human resources—and the multiple combinations of these—are not perfectly
divisible, the firm has incentives to push resource use up to is maximum capacity. It will
consequently seek innovations using the same resources. This explains why firms innovate
with similar technologies, that is to say, over path dependence trajectories.
On the other hand, since resource performance depends on how they are used, additional
resources or personnel (capabilities) could lead to different performances. Accordingly, the
firm has incentives to seek new uses of its resources by means of acquiring knowledge.
Given the fact that knowledge is also created during the process of production and
commercialization, there is a constant change in the level of resources, how they are
allocated, and how they perform.
Similar to the conclusion reached by the capability-based approach, the complexity and
dynamism of the competitive process (productive and commercial processes, in Penrose’s
terms), as well as the firm’s managerial capability to foresee changes and re-allocate
resources, explain the persistence of heterogeneous situations and trajectories at the firm
level. The observable result will be the firm’s actual capability, as an organization, to master
competences and resources in the search for a competitive advantage. As a result, the level
of resources is associated with path dependence while skills level and the learning processes
explain path creation. How the level of resources and capabilities are combined constitute
the firm’s strategy, which is a continuous process of decisions about acquisitions and
allocations of resources and capabilities and adjustments to these.
3.1.1.3. The knowledge-based approach
The knowledge-based approach is based on the acknowledgement of the impact of the
environment as a historical force shaping the firm’s characteristics—something less visible
in the other two approaches—and analyzes the observable outcome of innovation and
performance at the firm level (Lundvall, 2009; Lundvall et al., 2009a). The creation and
recombination of knowledge is at the center of the dynamic of capitalist competition,
Chapter 3: The strategy-based approach to firm behavior
- 92 -
which is shaped by the firm’s investments and its articulation (co-operation) with all other
organizations affecting the selection process (Metcalfe, 2002). According to this approach,
innovation is in fact new combinations of existing knowledge, and the firm’s competitive
position—how it takes advantage of the selection process—depends on its ability to learn
(Lundvall and Lam, 2007).
Knowledge is understood as a collective process, and how it is combined and transformed
into innovations depends on how the firm manages its assets and investments, as well as
the characteristics of the environment that surrounds it. For instance, the importance
assigned to formal training versus experience-based learning as a way of certifying
capabilities, and how the labor market promotes long-term employment versus individual
progress, will impact on the firm’s characteristics. They determine whether behaviors will
tend more towards the cooperative or the competitive, how and where individuals will
share their ideas, and the level of commitment between employees and employers
(Freeman, 1987; Arundel et al., 2007; Lundvall and Lam, 2007).
Basing their analysis on Penrose’s (1959) work, Lundvall and Lam (2007) acknowledge the
role of environmental conditions and identify three complementary functions of the firm:
to allocate resources (in the same sense as Nelson described the strategy), to exploit them
by entering into new activities, and to create new competences. In this sense, Penrose’s
resources are in fact different forms of knowledge the performance of which—in other
words, how far it is transformed into innovations—will depend on the firm’s competences.
In dynamic terms, the process of competence building allows the firm to face
environmental changes and to profit from knowledge. It follows that the strategic decisions
are those related to the processes of competence building and knowledge creation. In
practical terms, how internal competence building, external knowledge, and network
positioning are combined in order to transform knowledge into innovations is a matter of
strategic decision. As a result, the strategy of the firm is the set of choices that define how
knowledge will be created, combined, and applied in order to develop innovations.
As with the capability-based approach, firm organization is the observable outcome of the
strategic process of knowledge creation and transformation, although different
interpretations of this can be found within the knowledge-based approach. In some cases,
organization is viewed as the result of a specific managerial approach regarding work
organization and knowledge investments (Jensen et al., 2007; Lundvall and Lam, 2007). In
other cases, a firm’s organization responds to the sector it belongs to, which determines the
Chapter 3: The strategy-based approach to firm behavior
- 93 -
characteristics and sources of technological change (Pavitt, 1984; Malerba, 2004). In other
cases, similarly to Coase’s transaction costs (Coase, 1937), knowledge is treated as an input
that can be externally acquired or internally created, and firms organization is a matter of
strategic making versus buying decisions (Fraga et al., 2008; Frenz and Lambert, 2009;
Arundel et al., 2007).
In any case, and coming back to the issue of micro-heterogeneity, firms are different
because they are the result of a complex combination of knowledge and competences,
which, influenced by managerial capabilities and environmental features, will give rise to
unique trajectories and firm-specific characteristics. Within this approach, path dependence
accounts for the level of competences and environmental determinants, while path creation
is a matter of learning and competence building.
3.1.2. Consensus and limitations of the approaches
If one looks beyond the semantic differences between the three approaches presented in
Section 3.1.1, the elements and limitations they share become evident, especially as their
explicit recognition of each author’s contribution to the understanding of a specific
dimension of the firm behavior.
In terms of common features, the three approaches look for the same causal explanations
and rest on the same key assumptions. These are, firstly, that firm behavior can be
understand as the set of decisions regarding the resource allocation and the creation of
capabilities, subject to the characteristics of the environment; and secondly, that firms are
organizations the behavior of which is directed by a specific strategy aimed at growing and
surviving in a Schumpeterian competitive environment where innovation (knowledge) is
the key to success. Another common element is the recognition of how resources perform
differently according to the capabilities of the individuals that exploit them. In this sense,
the complex interaction between decisions (strategy), resources, and capabilities explains
why the productive structure is characterized by micro-heterogeneity.
The three approaches also share the theoretical explanation of the path-dependent nature
of the firm as well as possible alternative futures, that is, path creation. Path dependence
processes and environmental determinants account for the range of options the firm has.
Learning and accumulation processes, in turn, determine the spectrum of possible futures
the firm can aspire to, which will be subject to the firm’s ability to cope with the
environment, its levels of resources and capabilities, and the managerial skills required to
Chapter 3: The strategy-based approach to firm behavior
- 94 -
understand new options and adjust strategy accordingly. Surprisingly, while all three
approaches include an explanation of path creation, further analyses have focused almost
entirely on historical explanations of firm performance (path dependence) as if they were
passive to the trajectory that determines their competitive market position.
Two other implicit assumptions shared by the three approaches. The first is that
competition is mostly about innovation, which implies that strategic decisions are a matter
of how innovation will be carried out and neglecting the possible existence of other types
of strategic choices, many of them unsustainable in terms of long-run growth but profitable
in the short term. The second is that selection always leads to the disappearance of less
productive firms. The existence of these—even of entire obsolete productive sectors—are
consequently ignored as they are prone to disappear because of the “natural” evolution of
Schumpeterian creative destruction.
According to the approaches considered above, the usual explanation for the existence of
firms with low levels of productivity is based on matters of timing: sometimes creative
destruction takes longer (Teece and Augier, 2007). However, even if one assumes that the
least productive firms will disappear in a “couple of decades,”43 their existence at least
poses questions regarding the role of public policy and the institutions that affect Nelson’s
selection process.44
In this respect, there is a missing link between Nelson’s evolutionary theory of the firm and
his view of the NIS (Chapter 1.2). According to the NIS approach, the selection process is
affected by several organizations (Nelson and Dahlman, 1991) which, by definition, are not
neutral (Johnson, 1992). Moreover, each organization will probably have different
“selective” objectives, even if these are contradictory.45 As such, there is no a priori reason
to believe that this process will automatically lead firms to compete based on innovations
or will trigger learning processes capable of improving routines. Therefore, the selection
43 Keynes’ (1936) famous statement that in the long-term we are all dead seems more relevant than ever. 44 In 1994, Nelson asked himself the same question when he claimed that the impact of social and political forces have to be taken into account when analyzing the selection process (Nelson, 1994). However, he continued his arguments by focusing more on the institutional development necessary to support a new technology and not on the multiple possible organizations affecting firm evolution. At the firm level, much of this question remains unanswered in most of the theoretical contributions, while the selection process was associated with the market mechanisms that lead to the consolidation of a specific “dominant design”. 45 During Argentina’s Convertibility Plan, since exports were key to sustaining the economic model, the
importation of embodied technology was fostered as a way of gaining competitiveness; however, importing final products was the best option in terms of the cost and benefits of producing locally. As a result, firms faced incentives to import capital goods but also to substitute local production.
Chapter 3: The strategy-based approach to firm behavior
- 95 -
process within the NIS could alter Schumpeterian market competition by increasing micro-
heterogeneity.
There is also a missing link within the NIS approach regarding micro-heterogeneity, which
is a common characteristic of any capitalist economy. The micro-heterogeneity referred to
by the NIS approach and in the approaches reviewed here is understood as the variety that
allows the selection process to take place. According to these approaches, the diverse
situations at the micro level complement each other by compelling firms to compete and
improve their performance. In this way, the selection process leads to innovation and
innovation to increasing the society’s knowledge base and income. In this sense, it is
“good” micro-heterogeneity. However, the micro-heterogeneity is deeper in some systems:
it goes beyond the selection process and is instead concerned with the weakness of the
institutional set up (market regulations and enforcement, tax system, environmental
concerns, trust, etc.). This is usually a “bad” type of micro-heterogeneity, to the extent that
it allows firms to survive without innovation. The NIS where firms compete shapes the
first type of micro-heterogeneity and determines the characteristics and the depth of the
second.
Consequently, the relationship between the environment and firm behavior could be
redefined: firms take decisions regarding how to deal with the specific selection process
based on their characteristics and the NIS in which they operate. In doing so, they are
deciding how to compete (and not just how to make profits). In the terms of Fajnzylber
(1989), they are deciding how to reach a competitive advantage.
When analyzing the process of industrialization in Latin America, Fajnzylber (1989)
identifies two extreme theoretical types of competitive advantage: the spurious and the
genuine. At the country level, the former push production and exports on the basis of
periodical devaluations of the domestic currency and depressed salaries. With the latter,
competition is based on the quest for technological progress as a way of making
extraordinary profits and surviving the competition process. At the firm level, the spurious
competitive advantage refers to those firms which tend to defend or increase their market
share based on a “competitive” exchange rate, on the existence of captive demands, or on
cost minimization by means of low wages, labor rationalization, environmental
degradation, or market cannibalization. This type of advantage usually leads to short-term
private profits but it is not sustainable in the long run, unless firms have new spurious
Chapter 3: The strategy-based approach to firm behavior
- 96 -
reactions to changes in the environment (Ferraz et. al, 1996; Porta and Bonvecchi, 2003;
Lugones and Suárez, 2006; Porta, 2006; Ocampo, 2001, 2012).46
Of course, strictly speaking, a firm that is capable of reducing its costs by means of labor
rationalization is in fact a process innovator, and one that manages to take advantages of
grey areas in environmental regulations is an organizational innovator. However, this does
not seem to be the idea Schumpeter had in mind when he defined capitalist competition or
that any of the quoted authors refer to when they argue in favor of innovation and
technical progress.
The other type of competitive advantage Fajnzylber defines, genuine competitive
advantages, are based on technical progress and organizational change, which allow the
firm to stand out from competitors, either because its costs are lower or because
consumers are willing to pay higher prices for a differentiated product. Schumpeter,
Nelson, Lundvall, and most of the authors referred to seem to be thinking of this type of
competitive advantage when they argue that innovation leads to the firm making
extraordinary profits and the national system to grow.
As such, coming back to the micro-dynamics of innovation, this implies that Penrose’s
observations about resource allocation, Nelson’s strategy regarding structure and
capabilities, and Lundvall’s investments in knowledge assume that these decisions are about
how to innovate, while no attention is paid to other ways of competing. In this thesis I
assume that only when the firm decides to seek a genuine competitive advantage has it
chosen an innovation-based strategy.47 Given the environment and the firm’s resources and
capabilities, whether to seek innovations or not is a strategic decision that precedes choices
about how resources will be allocated or how much will be invested in knowledge.
Moreover, subject to their past learning processes and accumulation of capabilities, firms
can change their path dependence by means of changing the competitive strategy. As long
as innovation means knowledge creation, every time the firm chooses an innovative
strategy there is a window of opportunity for de-lock-in and path creation.
46 Continuing with the same example about incentives during Argentina’s Convertibility Plan, firms that based competition on the exchange rate become importers and made (high) short-term profits but then faced severe limitations to surviving the subsequent recession, given the decreased demand. They also found it hard to take advantage of the new growth period, given the loss of productive capabilities. 47 Of course, the fact that a firm has chosen an innovation-based strategy does not mean that it cannot “complement” the strategy with spurious advantages. Hybrid cases like this raise interesting research questions (e.g. how an artificially devaluated exchange rate creates incentives to export, which triggers learning-by-exporting processes, or how precarious employment schemes allow the firm to redirect resources to innovation), but they are beyond the scope of this thesis.
Chapter 3: The strategy-based approach to firm behavior
- 97 -
3.2. Strategic decisions and innovation dynamics
3.2.1. Innovative strategies and observable structures
Path dependence literature argues that the firm’s past decisions generate lock-in effects,
which narrow the range of opportunities they can take advantage of in the future (Nelson,
1995; Antonelli, 1997). These lock-in effects are not only technological but also structural.
Once the firm has selected a strategy, the structure will acquire a particular shape, including
the type of productive, human, commercial, and financial assets the firm chooses (Penrose,
1959; Nelson, 1991b). This strategic choice has emerged from the capabilities of those
responsible for making the choice but also from the firm’s general level of capabilities.
Once selected, the strategy will lead to a particular competitive position and will trigger
specific learning processes by means of feedbacks between the firm and the environment,
and within the different areas of the firm (Freeman, 1974; Nelson and Winter, 1982;
Lundvall, 1992; Langlois and Robertson, 1995).
The entire process—from the selection of the strategic choice to the final impact in terms
of profits, if it is successful—is surrounded by uncertainty, which decreases as events move
forward. As long as the strategy is transformed into concrete actions, the level of
uncertainty tends to decrease because the amount of information regarding potential
benefits and costs has increased, but also because the process itself triggers learning
processes that enhance the firm’s competences and its understanding of possible outcomes.
In the course of the process, the firm’s strategy will probably be adjusted based on the new
information and certainties (Ansoff, 1973).
Similar reasoning applies to the innovative process, although there are wider interpretations
for the possible outcome. On the one hand, the direct outcome of the process is the actual
innovation that allows the firm to obtain a differential rent. On the other hand, regardless
of the direct result—whether the firm has innovated or not—the process of innovating
triggers learning mechanisms that enhance human resource capabilities, generate new
knowledge that could trigger new innovation processes, and improve the firm’s
competences for future quests for innovation (Lall, 2001; Fagerberg, 2003; Jensen et al.,
2007; Lugones et al., 2007; Yoguel et al., 2011).
This dynamic process of strategic decisions, implementation, and adjustment explains why
Nelson (1991a, b) sustains that the evolution of the firm structure (its organization, in
Lundvall’s terms) can only by “dimly” foreseen. It also explains why Freeman and Soete
Chapter 3: The strategy-based approach to firm behavior
- 98 -
(1997) argue that those successful strategies which can be perfectly explained ex post can
hardly be defined ex ante, let alone by the firm that has carried them out. In other words,
the firm’s strategy can be observed by looking at the resulting structure from an ex-post
analytical perspective. It follows that, when looking backwards, innovation is the result of
strategic decisions and path dependence trajectories. When looking forwards, the strategic
decisions and the innovations they foster—and those they avoid—are the main causes of
path creation.
3.2.1.1. The evolutionary concept of the firm’s innovative strategy
So far, I have used Nelson’s definition of “strategy,” which refers to the set of decisions
the firm has made in order to face the selection process. According to this definition,
strategy is a conceptual construction that is very different from the idea of an ex-ante
detailed plan. It is the set of actions determined not only at the managerial level but also by
the reaction of the rest of the firm and the evolution of the environment. As this definition
of Nelson’s is based on the strategic management approach, some contributions from
management literature may help to clarify the key concept.48
Within management literature, any decision regarding the relationship between the firm and
the environment is considered a strategic choice. According to this conceptualization, the
firm’s strategy will be the set of decisions that relate to this issue and it will trigger
additional operative and administrative decisions. Given the complex and dynamic
characteristics of the environment and the firm (an aggregate of individuals and resources),
the strategy has to be clear enough to be a guideline for the organization but also flexible
enough to be adjusted while it is being implemented (Ansoff, 1973). As there are multiple
internal and external factors affecting the implementation of the strategy, analysts should
distinguish between what was intended and what was actually realized (Lynch, 2012). The
intended strategy is what the managers or owners want to do, the realized strategy is the
result of the combination between the deliberated strategy and the emergent one, which is,
once again, the result of the process of feedbacks and adjustments to the original strategy
(Mintzberg and Waters, 1985).
48 Although not reviewed in this paper due to the evolutionary approach it adopts, management studies have tackled the issue of the firm’s strategic behavior for long time, often based on the same Schumpeterian premises (Ansoff, 1973; Mintzberg, 1983; Porter, 1985, 1990; Drucker, 1964). Although the influence of this line of thinking can be observed in some innovation studies with an economic perspective, there is a historical lack of dialogue between the two disciplines, which has worked against the potential benefits of combining their approaches. Building bridges between the two disciplines seems an interesting field for future research.
Chapter 3: The strategy-based approach to firm behavior
- 99 -
In some extreme cases, firms have a clearly delimitated strategy with a high degree of
control over variables and low levels of uncertainty. In these cases, the realized strategy has
a greater proportion of deliberate strategic actions. In other cases, the exact opposite
position can be found: the initial strategy is just a set of general objectives, with little
control over the environment and the rest of the organization. In these cases, the final ex-
post observed strategy has a greater proportion of emergent factors, that is, of emergent
strategy. Large firms or firms operating in stable markets will probably have more
deliberate strategies, while small firms or firms within dynamic sectors will just have a set
of general goals and they will reach a more emergent strategy. In both cases, once the initial
set of decisions has been implemented, subsequent decisions will impact the strategy,
whether by reinforcing the original plan, redefining it, or reshaping it.
This is the concept of strategy applied in this research. It is the set of planned or emergent
decisions regarding how to face the selection process. This concept accounts for a
continuum between the initial goals, which are more or less planned, and the final results,
where each decision that affects the general goal constitutes a strategic choice. Of course, it
is an ex-post concept which assumes that the firm’s observed strategy is the result of
deliberate actions, unplanned events, and learning processes which allow the firm’s
behavior to be adjusted to deal with the environment and with its own organization. The
observable outcome is the firm’s competitive market position but also its structure,
resources, and capabilities (Nelson’s structure; and Nelson’s, Penrose’s, and Lundvall’s
definitions of resources and capabilities).
According to this definition, and coming back to the innovation process, when the firm
decides to face competition by means of innovation, then it has chosen an innovation-
based strategy. The observable feature of this strategy relies on the structure of the firm:
investments in innovation activities, allocation of human resources to these tasks,
interaction with external sources of knowledge, etc. In practical terms, investments towards
the creation, acquisition, and modification of knowledge are the main type of resource
allocation of an innovation-based strategy.
Resource allocation, whether is one-time event (such as a new machine) or a more long-
term project (such as an R&D lab), should account for the nature of the sought-after
innovation but also for how systematically the firm bases its competitive advantage on
innovations. The more structural the firm’s resource allocation, the greater the complexity
of the learning processes and the higher the probability of the firm getting involved in new
Chapter 3: The strategy-based approach to firm behavior
- 100 -
innovation projects. However, the more structural the allocation of resources, the higher
the lock-in effects the firm can face.
In this regard, two different approximations to the analysis of innovative strategy and the
consequent firm structure are worth reviewing: Freeman’s (1974) innovative strategies and
Pavitt’s (1984) taxonomies. On the one hand, from a historical perspective, Freeman’s
contribution was a critical response not only to the neoclassical free and costless
technology that could be introduced into any economic analysis as an external shock, but
also to the standard theory of the firm, in that it provides a deeper understanding of firms
as organizations with bounded rationality and fuzzy borders where uncertainty and multiple
objectives coexist. Pavitt’s taxonomy, on the other hand, was an empirical proof regarding
the different sources of technological change and the need to understand innovation at the
sectoral level, especially in terms of policy recommendations.
3.2.1.2. Freeman’s innovative strategies
In 1974, Freeman warned that firms behave differently when they are faced with
technology (Freeman, 1974). The author argues that, during a given period, firms adopt a
specific strategy, which consists of the combination of the resources and scientific and
technological skills the firm allocates to competing in the market (structure and capabilities,
in Nelson’s terms). Based on this definition and a large set of empirical studies, he
identifies six different “ideal types”49 of innovative strategies: offensive, defensive,
imitative, dependent, traditional, and opportunist.
Offensive firms are those that constantly try to move the technological frontier. Thus, their
innovative dynamic is based on the performance of science-based activities, from which
value is added, but also on a strong capacity for “educating and training” consumers and
suppliers—Lundvall’s user-producer interactions—as long as the goods it produces are
entirely new. In this case, the strategy is about radical innovations, the structure is based on
strong investments in endogenous creation of knowledge, and the capabilities are attached
to the performance of R&D, the way understanding demand is understood, and the
communication of the products.
Firms with a defensive strategy also based their structure on the performance of R&D
activities, but an innovative strategy aims to improve the first mover’s innovation and is not
49 The author refers to “ideal types of strategies” while acknowledging the complexity of reducing reality to a simple classification. However, he also recognizes the methodological power of “ideal types”, and it is in that sense that he proposes these strategies (Freeman, 1974, and Freeman and Soete, 1997).
Chapter 3: The strategy-based approach to firm behavior
- 101 -
necessarily associated with a movement of the technological frontier. The structure and
capabilities of these firms are oriented towards the identification of key innovations—
technological watch via patents—and improve them. They do not take on the cost of
training customers and suppliers although they have to invest in having a fluid interaction
with the different sources from which new innovations may originate, as well as the
flexibility required to quickly master and improve a new technology. In this sense, the
author highlights the importance of relationships with the S&T infrastructure and the
creation of connectivity capabilities.
Freeman defines imitative firms as having a defensive strategy in a developing country or
any other localization advantage that guarantees low costs or captive demand. This type of
innovative strategy leads to a structure based on engineering activities and technical
information, and the firm is focused on the quest for a superior competitive position based
on the acquisition of innovations developed elsewhere (by intra-firm technology transfers
or licenses) or constant cost reductions by means of process innovations. Capabilities, in
this case, are strongly related to engineering.
Dependent firms are similar to imitative ones in that they implement innovations
developed elsewhere as an adaptive response. This innovative strategy is found mainly
among medium and small suppliers of large firms, which are the ones innovating and
therefore demand changes to the supply chain. Since dependent firms have to be prepared
to quickly adapt their structure to external innovations, they have high engineering and
industrial design capabilities and almost nonexistent S&T capabilities.
Traditional strategies are present in those firms where the product remains the same or
only undergoes fashion changes. These firms base their survival on quick adaptation to
these fashion changes although, as expected, they are vulnerable to external technological
shocks. As with the previous strategy, their capabilities focus on improving the process and
quality assurance, with almost nonexistent investments in R&D. According to our
definition, this strategy is not necessarily an innovative one—whether it is will depend on
the magnitude of the novelty of the new product.50
Finally, there is opportunistic strategy. Freeman devotes only a few lines to this type of
behavior and defines it as those firms which can quickly identify short-term niches in
50 According to the Oslo Manual definition of “innovation” adopted in this thesis, not every packaging change can be counted as an innovation: only “significant” changes in the product, process, organizational, or commercial practices can be. See OECD (2005).
Chapter 3: The strategy-based approach to firm behavior
- 102 -
rapidly changing markets and exploit them by supplying a product or service which does
not require R&D but which no-one else will provide. Probably the most important feature
of this strategy is the figure of the entrepreneur and his or her ability to identify
opportunities. In this case, there is a vast range of strategic options so these behaviors
could be associated either with innovative or non-innovative strategies.
3.2.1.3. Pavitt’s sectoral taxonomy
Pavitt’s taxonomy is probably the most widely used classification of innovative behaviors
and, as was mentioned in Chapter 1.1, it is the basis of sectoral approaches to innovation.
While Freeman analyzes strategies at the firm level, Pavitt looks at the productive structure
and distinguishes four types of sectors, according to source of technological change (or the
source of knowledge): science-based, specialized suppliers, scale-intensive, and supplier-
dominated (Pavitt, 1984).51 The innovative strategy of the firm is, consequently, the
observable manifestation of how technological change occurs within the sector it belongs
to.
As with Freeman’s offensive and defensive strategies, in science-based sectors, firm
strategy is based on R&D performance to push the technological frontier. It is also based
on an intense use of the patent system and other legal methods for protecting innovations.
In contrast to Freeman’s strategies, science-based firms have also strong capabilities in
engineering as long as dynamic learning economies in production are among the main entry
barriers. Although the main source of knowledge is within the firm (the R&D lab), these
firms also present strong interactions with S&T institutions. Consequently, innovative
strategy is the same as Freeman’s offensive and defensive strategies, with similar structures
and similar capabilities.
Specialized suppliers and scale-intensive firms are two strategies where the source of
competitiveness lies in the productive process. In a sense, this is similar to Freeman’s
dependent and imitative strategies, although it is determined by the productive process and
not by a managerial strategic decision. Firms with a scale-intensive innovative strategy have
strong engineering capabilities and less R&D competences as long as their competitive
advantage lies in the low cost of the final product. Improvements and extraordinary profits
are made up of the interaction between the incorporation of technological change through
51 The same as Freeman, Pavitt´s taxonomies are based on several decades of empirical research and teorethical analysis over the productive dynamic of firms, in both cases, with a bias towards the British productive structure.
Chapter 3: The strategy-based approach to firm behavior
- 103 -
suppliers (basically, capital goods) and investments in in-house competences to adapt and
improve the productive process. Specialized suppliers, on the contrary, are capable of
responding quickly to their customers’ needs—to a certain extent, similarly to Freeman’s
dependent strategy—although these firms are more focused on the performance of the
goods than on a price-based competitive position. In this case, technical change is mainly
triggered by customers requirements, which demand strong capabilities in product
development and design.
Finally, supplier-dominated sectors are similar to Freeman’s traditional strategy to the
extent that the product offered by these firms tends to remain the same over time.
However, Pavitt’s classification adds the fact that in those sectors where the product
remains the same, price competition tends to be intense since consumers are price-
sensitive. In these sectors, technological change is incorporated through embodied
technology (capital goods and raw materials) while disembodied knowledge is incorporated
through external hiring—S&T services. Therefore, the process improvement is at the
center of the innovative strategy and the structure is based on investments in machinery
acquisition, probably with lower capability levels.
3.2.2. The static perspective of the strategic process
More than 20 years after these pioneering classifications of innovative strategies, and
thanks to the multiplication of available information, several studies have been performed
based on these strategies.52 These studies have significantly contributed to the
understanding of micro-heterogeneity and the complex nature of the innovation process.
They have also provided evidence to support Penrose’s ideas regarding the different
performance of resources due to the different capabilities for using them; Lundvall’s
arguments regarding the different ways of creating knowledge and transforming it into
innovations; and Pavitt’s and Freeman’s ideas regarding the source of technical change
within the firm. All of them reach the same conclusion: firms carry out different strategies,
which lead them to reach a specific innovative structure and different economic
performance.
However, like the contributions of earlier authors such as Nelson, Penrose, and Pavitt,
these subsequent studies all share a static perspective of the strategic process of innovation.
They can explain why a firm performs better but they cannot shed light on how firms react
52 See, for instance, Leiponen and Drejer (2007), Clausen et al. (2011), Srholec and Verspagen (2012), Frenz and Lambert (2009), among others. These and other studies will be analyzed in depth in Chapter 4.
Chapter 3: The strategy-based approach to firm behavior
- 104 -
to that improved performance or, in Nelson’s terms, how the selection process triggers
learning processes that improve routines. They can explain how the firm’s innovative
strategy is based on a high level of capabilities but they cannot explain how learning
processes that arise from the combination of capabilities and resources feed back into
strategic decisions or, in Lundvall’s terms, how firms learn and transform knowledge into
innovations. In other words, these approaches cannot explain how strategies evolve or to
what extent firms can master path creation.
An aspect that it is only marginally discussed in the approaches reviewed in this chapter is
how firms react to different environmental incentives in such a way that they are in fact
radically modifying their competitive advantage. In the case of the capability-based
approach, there is no clear explanation about how the interaction between capabilities and
learning leads to the disappearance of some routines and the installation of others. In the
case of the resource-based approach, no explanation is provided regarding how the
knowledge created during the productive process leads to a realignment of strategic
objectives, resources, and capabilities. Within the knowledge-based approach, only a few
attempts have been made to explain how a firm manages to escape from the environmental
determinants. In short, although all the approaches and taxonomies base their explanations
on the dynamic process of innovating and competing, they fell short on explaining the
dynamic process of strategic change at the firm level.
3.3. The strategy-based approach: towards a dynamic theory of the
innovative firm
Following what has been discussed in this chapter so far, firms take decisions based on
their resources and capabilities and the environment in which they operate. When these
decisions are related to how the firm will face the selection process, then they are taking
strategic decisions. When these strategic decisions are related to the quest for a genuine
competitive advantage, then the firm is following an innovation-based strategy. Thus, the
firm’s innovative strategy is the set of all those strategic choices aimed at developing a new
or significantly improved product, process, or organizational practice—in other words, an
innovation—and it is just one of many possible strategies the firm can carry out to survive
and grow in the market.
From an ex-ante perspective, the innovative strategy—indeed, any strategy—lies
somewhere between a carefully planned set of actions and a general idea of the firm’s goals.
Chapter 3: The strategy-based approach to firm behavior
- 105 -
When it is observed ex-post, the strategy is the particular combination of deliberate actions
and emergent reactions. It is a continuum of decisions, implementations, and adjustments
and it goes beyond a specific innovation project.
At a given point in time, we can make an analytical cut to observe and evaluate the
innovative strategy in terms of results and impacts. The result of the strategy can be
acknowledged by looking at the innovations the firm has or has not achieved, their scope,
scale, degree of novelty, etc. The strategy has a two-fold impact: static and dynamic. The
static impact is the firm’s final competitive position, and a successful strategy will improve
the firm’s market position. The dynamic impact of the strategy, however, is more difficult
to judge. It is the firm’s remaining structure and new level of capabilities, which are two of
the three elements that condition the future set of strategic decisions. In this case, there are
no successful or unsuccessful strategies, but a set of elements that determine the
subsequent processes of strategic decision, implementation, and adjustment. This is
precisely the link between the firm’s path dependence (the impact of previous strategies)
and path creation (how the firm takes advantage of that) and how it manages to change its
future.
The third element, the environment, will also impact the firm’s path creation and will be
included in strategy-related considerations. Therefore, the range of possible decisions the
firm can take is the result of the combination of exogenous opportunities and the firm’s
capabilities to understand and react to them. At the more aggregate level, the
macroeconomic situation, the institutional set-up, and the NIS are the most obvious
environmental determinants. 53 At a more meso level, environmental determinants are given
by the position the firm has in the network it belongs to: the sectoral belonging of the firm,
its capital ownership, its size, its place in the national (or international) value chain, the
nature of the local innovation system, the scientific and technological breakthroughs, etc.
Once the decisions have been taken, the firm will allocate resources to pursue its chosen
strategy, which will in turn shape its structure and impact the relationship it has with the
environment (a better competitive position, the market share, new linkages with other
organizations, etc.). In a sense, the impact of the innovative strategy can be also
acknowledged by looking at the new relationship with the other NIS organizations and the
firm’s network positioning.
53 See Chapter 1.2.2.3 for a more detailed discussion of the environment and how it is defined in this thesis.
Chapter 3: The strategy-based approach to firm behavior
- 106 -
From a dynamic perspective, from one moment to the next, the range of options the firm
considers when defining its strategy can change, whether because of changes to the
environment, to the firm’s capabilities or resources, or a combination of these aspects. In
contrast, since the firm’s dynamic is subject to path dependence trajectories, the specific
choices the firm has made also trigger lock-in processes that narrow the range of
opportunities in the subsequent period. Of course, firms can also make bad choices, fail at
what they were doing, or simply make mistakes at reading the opportunities. In these cases,
although learning-by-mistake processes will impact firms’ capabilities, opportunities in the
subsequent period will probably be reduced, especially because the firm is in a worse
competitive position. In any case, the modified capabilities will lead to strategic decisions
being reviewed and the strategy adjusted before the process starts all over again. This
process takes place because firms learn. It is not the passage from one equilibrium to
another one but an evolving path of learning, adaptation, and competence building.
This recurrent process of decision-making and adjusting is the result of firms’ learning
capabilities and the existence of mechanisms that allow for knowledge accumulation—
routines, in Nelson’s terms; competence building, in Lundvall’s terms; and resources, in
Penrose’s terms. Two forces converge during the process of feedbacks to the strategic
decision process: the learning process within the firm and changes in (and feedbacks from)
the environment. Learning processes will enhance and shape the firm’s capabilities to face
the selection process and take new decisions. Changes in the environment will shape the
firm’s range of possible (profitable) strategies. The greater the firm’s capabilities for dealing
with the environment, the greater its ability to take advantage of environmental
opportunities or to avoid being constrained by them. Thus, the firm’s learning capability is
what explains why successful innovative firms can be found in unsuccessful environments.
3.3.1. Learning processes and interactions with the environment
Once the strategy has been set in motion, feedback processes will be triggered within the
firm and between the firm and its environment. Given the level of capabilities, these
feedbacks will lead to learning and accumulation processes, which will enable competence
building. Within the firm, there are two extreme types of learning process: those based on
scientific and technical knowledge and those based on experience (Jensen et al., 2007).
Between the firm and the environment, learning processes arise from the absorption of
external knowledge.
Chapter 3: The strategy-based approach to firm behavior
- 107 -
As with Freeman’s (1974) ideal strategy types, Lundvall’s modes of learning and innovation
are extreme theoretical separations, but the two could take place simultaneously in any
firm, complementing each other (Jensen et al., 2007). Like Penrose’s (1959) interpretation
of the productive and commercialization process as generators of new resources,
Lundvall’s modes also explain why regardless of the innovative result, the innovation
process is the key to understanding how the firm learns and improves its capabilities.54
Lundvall (2009) refers to learning as both the process of adaptation and the process of
competence building (capability increasing). The process of competence building can take
place on-the-job through doing, using, and interacting (the DUI mode) and increases
individuals’ competences to cope with theoretical and practical problems by means of
producing “know-how” and “know-who”. The knowledge involved in this mode is
basically tacit, although it can be codified. Since the firm is more than the aggregate of
individuals, the individuals’ competence–building process will improve the level of the
firm’s competences more than proportionally. As a result, the greater the firm’s
competences, the greater its capabilities to face new problems and adapt itself.
The science, technology, and innovation mode of learning (STI mode) arises from the
scientific and technological activities developed within the firm and creates “knowing-why”
knowledge. In this case, resources are allocated to the quest for new basic knowledge, new
combinations of already existent knowledge and, eventually, innovations. Within this mode,
the formal training of individuals is a key determinant of the competence building process.
Learning processes within the STI mode tend to lead to the creation of codified
knowledge, which is less firm-specific than the DUI mode. In this sense, firms are prone to
taking strategic and legal measures to control the process and, especially, its results.
However, as with the DUI mode, regardless of the result, the process will increase the
competences of the individuals involved and, consequently, the firm’s competences to deal
with new problems or to search for new knowledge (Jensen et al., 2007; Lundvall, 2009).
Regarding the learning processes that arise from the interaction between the firm and the
environment, there is, on the one hand, the distinction made in Chapter 1.2.2.4 between
the intended or unintended nature of linkages. There are some interactions that the firm
has explicitly sought through its objectives and even the commitment of resources. These
cooperation agreements are, by definition, ways of sharing and creating knowledge and the
54 It is worth noting that Lundvall’s modes of learning and innovation are also a critical response to the excessive attention to research and development to the detriment of other sources of knowledge and innovation, which are equally important in terms of the firm’s growth and survival.
Chapter 3: The strategy-based approach to firm behavior
- 108 -
same STI and DUI modes of learning applies. The second type of linkage are those that
arise from the process of commercialization and competition in the market and can be
classified into three types: market-based, knowledge-based, and institutional interactions.
Market-based relationships and how they impact the firm’s competences has been widely
analyzed in the literature (Cohen and Levinthal, 1990; Lundvall, 1992; Freeman and Soete,
1997; Tether, 2000; Tether and Swann, 2003). The process of commercialization leads the
firm to interact with its consumers and competitors and it creates and circulates
information about tastes and preferences, other firms’ strategies, and potential new
suppliers. Although the transformation of this information into useful knowledge will
depend on the firm’s capabilities, its circulation is inevitable.55
Knowledge-based interactions have been less analyzed in the literature, since they usually
do not include formal agreements and therefore information is scarce. They are those
linkages that aim to incorporate knowledge into the firm but not in the form of a
cooperation agreement, or that explicitly aim to develop an innovation on the basis of the
firm’s regular activities. This is the case with hiring or training employees and then
interacting with the educational system, or certifying or testing products and processes and
therefore interacting with S&T institutions. These interactions also trigger learning
processes which improve the firm’s capabilities, at least, by establishing initial connections
for more formal and complex future interactions.
Finally, there are the learning processes that arise from the institutional-based interactions,
which have been studied even less within innovation theory at the firm level even though it
is a matter of central importance within the NIS approach. As was mentioned before, the
selection process is affected by several institutions that regulate—or at least affect—its
dynamic. When facing the competition process, the firm will have to deal with the tax
system, labor regulation, the exchange rate, inflation, the environmental concerns of the
population, etc. There is a wide range of dimensions that affect the process of producing
and selling, which are formalized in organization through specific strategies, structures, and
capabilities. Depending on the institution it is facing, the firm will have to solve different
problems and then DUI—and sometimes STI—learning processes will be triggered.
55 Coming back to the example of firm’s behavior during the Convertibility Plan, the commercialization of imported products led the firm to know the level of price-sensitiveness of consumer and the value given to quality of the products; it also led the firm to interact with external markets and found new goods to commercialize and how to exploit the commercial chain as an entry barrier given the intensification of the competition.
Chapter 3: The strategy-based approach to firm behavior
- 109 -
Summing up, once the strategy has begun to be implemented, in addition to the impact of
the innovation process, production and competition will trigger additional learning
processes that will also enhance firm’s capabilities. These enhanced capabilities will not
only feed back into the strategic decision process but they may also lead to a better
performance of the resources. The strategy will then be adjusted, which will in turn impact
the structure and the enhanced level of capabilities. The implementation of the renewed
strategy will start this process of feedback and accumulation all over again.
3.3.2. Innovative strategies and competence building
Within an innovative strategy, the search for knowledge is an explicit component of the
firm’s structure, and human, financial, and productive resources are allocated to the
creation of valuable knowledge. Given the explicit quest for knowledge, the feedback and
accumulation processes triggered by this type of strategy have a greater impact on the
firm’s competences. On the one hand, since the firm has allocated resources to enhancing
its innovative capabilities, the firm’s general level of competences will be directly increased.
On the other hand, since the objective of this process is the quest for a new combination
of knowledge, the complexity of the information that will circulate during the process is
greater, and so the probability of developing new knowledge is also greater. As a
consequence, the feedback and accumulation processes triggered by the circulation of more
complex knowledge, coupled with the enhanced capabilities, will lead to a higher level of
competence building. This explains why, regardless of the result, investments in innovation
activities are a means to increasing firms’ capabilities. This also explains why innovative
firms deal better with the environment.
Of course, different types of strategies will lead to different learning processes. The
learning processes of a firm that has acquired a new machine in order to improve its
productive process (innovative strategy) will probably have higher impacts on its
capabilities compared to a firm that has not (non-innovative strategy). However, these
impacts will probably be lower than in a firm which also trained its employees to use the
machine in question. In this sense, the combination of formalized knowledge (STI mode)
with experienced-based knowledge (DUI mode) and the combination of knowledge created
exogenously (machines) and endogenously (engineering, R&D) have been proven to
maximize the impact of any innovative strategy on the process of competence building
(Yoguel, 2000; Jensen et al., 2007; Lugones et al., 2007).
Chapter 3: The strategy-based approach to firm behavior
- 110 -
It follows that, given the multiple issues affecting firm strategy and the multiple ways of
creating and combining knowledge, there is no a priori reason to expect two firms to
behave the same or to reach the same results, let alone the possibility of anticipating the
impact of the strategy in terms of resources and capabilities. However, the learning
processes triggered during the innovation process explain why past innovations impact
future ones. That is, they explain why we should expect innovative firms to innovate
persistently, and why successful firms tend to sustain their position while unsuccessful
firms tend to disappear from the market.
3.3.2.1. Persistence of innovation
The concept of “persistence” is used to define the correlation between past and present
innovations; that is, the way the firm’s innovative dynamic impacts the possibility of
obtaining results over time. It is based on the Schumpeterian idea of creative accumulation
(Schumpeter, 1942). According to Malerba et al. (1997), innovation persistence can be
understood as the serial correlation between past and present innovations and the statistical
demonstration of the binomial accumulation-feedback which emerges from the firm’s
interaction with the selection process. Under the proposed strategy-based approach,
innovation persistence is the observable result of the above-described interactive and
accumulative processes of learning triggered by the succession of successful innovative
strategies.
From a theoretical perspective, three complementary approaches to the phenomenon of
persistence can be found: one that centers the explanation on evolutionist path
dependence; one that sustains that persistence is the result of virtuous accumulation cycles;
and one where the focus is on the market power reached by firms that achieved
innovations. In all three cases, the key to understanding the phenomenon of persistence is
the acknowledgement of learning processes and dynamic scale economies—success breeds
success—and the dimensioning of its impact on the market. The differences between the
approaches lie in the determinants of that creative accumulation, the specific set of
feedbacks that allow it, and the extent to which accumulation and feedbacks determine the
firm’s trajectory in the market.
Among the precursors of path dependence approach is Antonelli (1997), who argues that
the firm faces non-ergodic path dependence processes which shape its future possibilities.
In this sense, the author stresses the combination of irreversibility and indivisibility, in a
Chapter 3: The strategy-based approach to firm behavior
- 111 -
context of structural actions that affect the firm, limiting its options. In an oversimplified
sense, in a particular space and time, past decisions generate sunk costs in terms of
resources (irreversibility) and set the margin to obtain scale economies (indivisibility). Both
aspects involve the opportunity costs of new decisions, which are weighed when the firm
decides how to face competition. Spatial and geographical grounding also contribute to
shaping the firm’s characteristics, thus reinforcing the existence of path dependence
processes.
Applying this to the firm’s innovative strategy, Antonelli (1997) argues that the
accumulation of events within the firm and its relationship with the environment
(multidimensional space) determine the set of possible decisions. Consequently, the firm
may modify its innovative path, but only from a narrow range of options. In turn, these
development paths are conditioned and condition the endogenous capabilities to develop
innovations. In this sense, the author highlights the learning processes arising from
innovation activities and the set of incentives emanating from the firm’s particular
economic situation (stage of growth or recessionary phase) as a source of explanation for
the phenomenon of persistence. Thus, the development of innovations in the past
contributes to competence building and generates opportunity costs in the present,
increasing the odds of deciding to tale on another innovative project, which obviously
affects the likelihood of actually reaching innovations.
The analysis of persistence in terms of virtuous cycles of accumulation is based on Nelson
and Winter’s (1982) explanations of the firm’s competitive dynamic. For these authors,
persistence emerges from the generation of feedbacks between past innovations, present
investments (allocation of resources), and future innovations. Based on their idea of
routines, Nelson and Winter argue that the decision process that leads to the innovation is
in fact a standard behavior, which, if successful, will be repeated. It is precisely the
persistence of routines which impacts the innovative features of the firm, either by
developing innovative projects, or by not engaging in them. When the firm succeeds in its
innovative strategy (that is, when it achieves an innovation), it stands out from the
competition and obtains a monopoly rent, which improves its financial situation,
generating surpluses that can be reinvested in the quest for new innovations. This success
also creates entry barriers that allow the firm to extend its appropriation of that extra
income. In other words, success not only breeds success but also contributes to
undermining competitors’ chances of succeeding.
Chapter 3: The strategy-based approach to firm behavior
- 112 -
The market power approach can be found in the work of Phillips (1971), Mansfield (1962),
and Geroski et al. (1997), among others. Those who ascribe to this approach argue that to
the extent that a firm becomes innovative, it achieves a greater market power that allows it
to obtain an extraordinary income (increase the level of resources). Thus, past innovations
allow future ones to be financed. In the reverse situation, given the nature of innovation,
those firms that cannot generate sufficient surpluses to fund future innovations face major
obstacles to developing an innovative project, or higher costs given by the differential
interest rate arising from the additional risks of such projects.
Despite the different focus of each of the arguments on persistence described above, all
include the financial impact, the effect on the market power and the accumulation of skills
derived from innovations. All of them define persistence as the result of a dynamic and
heterogeneous process, conditioned by the existence of thresholds (capabilities and
resources), which, given the firm’s strategy, generates positive feedbacks for the firm
(learning processes, accumulation of results), but negative ones for the competition (entry
barriers, additional costs of technological catching-up). They also share the underlying idea
that the environment does not change. These approaches assume that what the firm did on
the past is useful for the things the firm has to deal with in the present. This matter opens
up a new question about persistence: what happens when the firm is innovating in an
unstable environment?
3.3.3. The new process of strategic decision and the impact of the environment
What if the change in the environment is so deep that the path dependence of the firm
works against its survival? What if past innovations are no longer a source of competitive
advantage? What if the accumulated competences of the firm are no longer useful? The
literature about persistence assumes that some firms are prone to disappearing as a result of
the Schumpeterian process of creative destruction or that they are casualties of the well-
known economic cycle. Reality has proved to be somewhat more heterogeneous. Although
some firms disappear during the Schumpeterian competition process and others during
economic downturns, some firms manage to survive both situations.
The limits of this explanation about the firm’s innovative dynamic lie in the scarce
attention it has paid to the firm’s capabilities for coping with the environment. Persistence
literature cannot explain how the dynamic of the innovation system triggers learning
processes that some firms manage to capitalize in the form of better capabilities for coping
Chapter 3: The strategy-based approach to firm behavior
- 113 -
with the specificities of the environment, which not necessarily are transformed into
innovations.
As was explained above, firms take decisions based on the environment, and the set of the
decisions regarding how to seek an innovation-based competitive advantage constitutes the
firm’s innovative strategy. If the firm has chosen an innovation-based strategy then it will
face different learning and accumulation processes than its non-innovative competitors,
and depending on the specific type of innovative strategy and the consequent result, it will
face even more virtuous learning processes than other innovative firms.
Throughout this process, there will be feedbacks to the firm’s strategy that will allow it to
be adjusted. However, these feedbacks will come not only from the implementation of the
strategy but also from changes to the environment. The most obvious source of changes in
the environment is, of course, the dynamic of other organizations. Not only are the firm’s
competitors following a specific strategy, but so are all the other organizations affecting the
selection process. If a media company is following a concentration strategy and rules
regarding media concentration have changed, then the firm will have to adjust its strategy
and allocate resources either to look for other possible strategies or to legally fight the
regulation.
A similar reasoning can be applied to the impact of the evolution of the economic
environment: variations in the exchange rate, a new macro-economic policy, a new public
intervention, or the deregulation of a market are all possible environmental situations that
will lead the firm to adjust its strategy. They are all feedbacks that carry information to the
decision-making process. The deeper the environmental changes, the more radical the
strategic adjustments will be. How accurately these feedbacks are interpreted and
transformed into a new strategy will depend on the firm’s dynamic capabilities. Similarly,
how fast the firm reacts by creating a new strategy to face the competition process
differently depends on the firm’s ordinal capabilities and its resource levels. The observable
result in this case is usually referred to as “the survival of the fittest.”
Every change in the environment will feed back into the firm’s strategic decision process.
As a result, when facing a major change, the firm will rethink its strategy and take new
decisions regarding how to seek out competitive advantage, including whether to follow an
innovative strategy or not. This is so because these environmental changes affect the
competition process (the selection environment) and modify the set of incentives. The
Chapter 3: The strategy-based approach to firm behavior
- 114 -
firm’s trajectory in terms of previous learning processes will determine its level of dynamic
capabilities and therefore its margin for mastering its own path creation. Since innovative
firms have gone through more complex learning processes and have also allocated more
resources to increasing their level of capabilities, they will probably be in a better position
to cope with environmental changes.
Summing up, the lock-in effect of path dependence will lead to the disappearance of some
firms when faced by changes in the environment. Other firms, however, will be in a better
position to successfully take this on and even to set up de-lock-in strategies. These firms
will change with the environment. In this case, path dependence has to do with prior
learning processes where the accumulation of competences improved firms’ capabilities to
learn, which in turn feed backed their dynamic capabilities to grow and succeed in both the
old and the new market. Therefore, persistence of innovation is about path creation, the
increase in firm’s capabilities, and how well they have learnt to deal with the environment.
3.3.1. Environmental shocks and resilience
There is a second type of environmental impact, which although part of capitalist
competition, are one-time, less predictable events, usually referred as external shocks. This
is the case with natural disasters (such as Hurricane Katrina), major technological
breakthroughs (the ICT revolution), an economic crisis (the 2008 financial crisis), etc.
From the point of view of the firm, national economic recessions can act as these deep
external shocks. To explain the impact of these events, evolutionary theory has appealed to
another biological metaphor: the concept of resilience.
Although an analysis of isolated shocks is not part of this research, a brief discussion of the
concept of resilience and how it can contribute to explaining firms’ innovative behavior in
an unstable environment is worthwhile, as it can contribute to clarifying the difference
between the definition of an “unstable environment” used here and the traditional concept
of “economic crisis.”
During the last decade, the succession of natural disasters and, more recently, the
international economic crisis have led to the proliferation of literature about the ability of a
system to deal with isolated exogenous shocks—that is, its resilience. The concept of
resilience, which originated within biology and psychology literature, has been applied to
economic literature basically by economic geographers, but also by some analysts of the
Chapter 3: The strategy-based approach to firm behavior
- 115 -
firm from both the management and evolutionary neo-Schumpeterian approaches.56 As is
usually the case with biological concepts applied as metaphors for economic dynamics,
resilience is not a precise concept and different definitions can be found for it. What is
more, given its recent application to economic approaches, different classifications of the
different definitions are present in contemporaneous scientific works (Hassink, 2010;
Martin, 2011). In simplified terms, the concept of resilience has been applied to economic
systems in four different ways: a system’s ability to return to pre-shock levels (engineering
resilience); a system’s resistance to external shocks (ecological resilience); a system’s greater
adaptability and capacity for renewal (evolutionary resilience) (Martin, 2011); and, within
management literature, a firm’s ability to grow from disruptions (organizational resilience)
(Moran and Tame, 2012; Davies, 2011).
Although external shocks are a useful concept, resilience in this sense cannot be used in an
analysis of the firm’s innovative behavior without several considerations that ultimately
undermine the relevance of the concept when trying to answer the questions that guide this
research. Specifically, the considerations have to do with the underlying assumption of
equilibrium, the strict delimitation between the system and the rest of the world, and the
treatment of the source of disruption as an exogenous isolated event.
Whether resilience refers to the recovery of pre-shock levels (the engineering approach) or
the achievement of a new post-shock situation (the ecological approach), there is an
underlying idea of equilibrium. Since the idea of accumulation and change lies at the core
of the explanation of firm behavior, there are no expected equilibriums but rather evolving
paths. At the same time, this equilibrium-based approach leads resilience literature to focus
on pre- and post- situations while paying little attention to the process that takes the system
from one level to another (e.g. the processes of recession and growth), the characteristics
of the two levels, and the interactive relationship between the system being analyzed and
the external system where the shock originated. As such, it is a useful conceptual tool for
analyzing impacts but is inadequate to answer questions regarding what happens to the
system during the process and why.57
The second limitation has to do with the definition of the system under study. The concept
of resilience is based in the idea of an external shock, where positive and negative impacts
56 See for instance Martin (2011), Moran and Tame (2012), and Østergaard and Holm (2010). 57 To some extent, the evolutionary and managerial approach to resilience has tackled this subject but focuses on the endogenous determinants of resilience capability with scarce consideration of the interactive nature between the firm and structural exogenous determinants. See for instance Moran and Tame (2012).
Chapter 3: The strategy-based approach to firm behavior
- 116 -
on the organizations taking part of the system are analyzed afterwards.58 In all cases, an
exogenous event, which could be recurrent but which is not regular, is assumed to disrupt
the system—the firm, the local/regional/national economy, the productive cluster, etc. In
the case of this thesis, the system could be the firm and it could also be the NIS, and as
such, the subject of study, what internal and external mean is not clear. Moreover, it is
precisely the interaction between the firm, the NIS, and the environment, the lack of a clear
distinction between actions from the environment and reactions on the part of the firm,
that determine the research questions of this thesis.
Finally, the third limitation—and probably the most obvious one—is the fact that this
thesis is about unstable environments, which means recurrent disruptions. If disruptions
are a regular feature of the system, then the existence of endogenous processes triggering
those disruptions and the existence of feedbacks and accumulation mechanisms that allow
the system and the organizations to learn from past events are key elements to
understanding actions and reactions. As a consequence, although the concept of resilience
can be used to characterize the short-term impact of a change in the rules of the game, it is
less suitable to account for the nature of learning and path dependence.
To conclude, although resilience constitutes a useful concept through which to approach
the impact of external events, it is in the early stages of development, with fuzzy borders
and generic applications. It has helped analysts to approach the impact of some events that
shocked several economies all over the world and to think about the limitations of those
approaches based on the idea of stability. It has also led to a new search for answers
regarding how to recover the growth path and be prepared for the downturn of
Schumpeterian capitalist competition. Clarifying this concept and approaching the study of
resilience from a dynamic and complex perspective seems a promising line for future of
research, which exceeds the objective of the present thesis.
3.4. Preliminary conclusions
The objective of this chapter was to provide a theoretical approximation to the innovative
dynamic of the firm, capable of accounting for both path dependence and the path
creation. In doing so, it aimed to contribute by zooming in on certain aspects of the NIS
approach in order to provide a more complex interpretation of the system’s impact,
beyond the impacts already described in terms of historical processes and idiosyncrasies.
58 See for instance Østergaard and Park, 2013.
Chapter 3: The strategy-based approach to firm behavior
- 117 -
The objective was to explain why successful firms can be found in unsuccessful
environments. The answer to this question lies in the firm’s learning capabilities and the
feedbacks and accumulation processes that arise from the interaction between the firm and
the environment, and the path of strategic decisions regarding how to face the selection
process. To explain these ideas, the strategy-based approach to firm behavior was
proposed.
In the context of capitalist competition, firms take decisions regarding how to face the
selection process. These decisions are both cause and consequence of the firm’s capabilities
and resources and constitute its strategy. When the quest for a competitive advantage is
based on the quest for innovation, then the firm is following an innovative strategy, and
resources and capabilities will be allocated to the creation and application of valuable
knowledge. Learning processes that arise by implementing the strategy will enhance and
shape the capabilities of the firm to face the selection process and the strategy will be
adjusted. Regardless of the results, these firms will face more complex learning processes
and therefore greater probabilities of developing capabilities and accumulating resources
that enable them to stand out from the competition and make extraordinary profits in the
future. In other words, they will be able to persist with innovation, not because of past
results but because of past behaviors. Changes in the environment will also impact strategy
and will trigger new adjustments. Since firms are the results of the interaction between
strategy, capabilities, and resources, there is no a priori reason to expect two firms to
behave the same or to achieve the same results, except within the neoclassical, purely
market-determined selection process. As a consequence, the productive structure will be
compounded by a set of heterogeneous organizations.
In an unstable environment, the selection process is weak and the institutional set up
allows multiple forms of competition (not always based on innovation). As a consequence,
some low-productivity firms will remain. In this case, firms may be able to survive without
innovating. This feature increases the heterogeneity of the productive structure and
explains why incentives to seek technological and organizational improvements are not an
automatic feature of capitalism. They are the result of the environment in which the firm
operates and the level of the firm’s capabilities of dealing with it. The higher the firm’s
capabilities the greater its possibilities of taking advantage of opportunities and avoiding
the negative impacts of the incentives that emanate from the environment. That is why
high-performance and dynamic innovative firms can be found in unstable environments.
- 118 -
Chapter 4: Towards a conceptual framework
for innovative strategies
Introduction
The aim of this chapter is to analyze how the strategy-based approach can complement the
literature about firm’ behavior and performance by proposing a classification criterion to
improve our understanding of the micro-dynamic of innovation. The main goal is to
discuss how innovative strategies can be translated into a model capable of identifying and
characterizing them. A critical review of the literature with empirical evidence is the first
step towards this objective.
Within innovation studies, three main contradictions arise when trying to test micro-
dynamics. Firstly, in order to test heterogeneity, heterogeneity must be controlled. Any
classification at the firm level will be arbitrary and will probably overlook some dimensions
that could potentially explain some of the variety of situations that can be found in a
specific environment (Freeman, 1974). Moreover, one classification could be relevant for a
specific innovation system during a specific period but not at another time or in another
place. In this sense, classifications are methodological tools; they are not universal
(ahistorical) axioms but analytical schemes for finding causal explanations. As such,
classifications should arise from the theoretical analysis of reality and should be contrasted
with reality afterwards. Therefore, the classification I will propose is one possible way of
studying the micro-heterogeneity of the productive structure of an innovation system. It is
not intended to account for all possible sources of micro-heterogeneity, but to improve our
comprehension of the variety within the system and to complement other analytical
classifications such as the firm’s sector, size, or location.
The second contradiction arises from the fact that to test a process based on feedbacks,
feedbacks have to be avoided. The assumption of exogeneity between the dependent and
explanatory variables is one of the most common restrictions on the micro-analysis of
innovation and is the source of many inconsistencies in estimations. In recent decades,
many econometric and software tools have been developed to “solve” endogeneity
problems.
Chapter 4: Towards a conceptual framework for innovative strategies
- 119 -
All the same, some of these problems remain unsolved. The question here is what to solve
the problem for when feedbacks and accumulation are the basis for the firm’s innovation
process. As such, a dynamic analysis will be proposed in an attempt to include the impact
of path—dependence, feedbacks, and accumulation. The possibility of a change in the
firm’s behavior will also be allowed for. By doing this, the interactive nature of the firm’s
strategic decision process can be analyzed without violating the assumptions about the
exogeneity of variables.
Thirdly, to test the complex dynamic of the innovation process, complexity must be
simplified. Except for the well—proven structural variables that explain part of the firm’s
dynamic (size, sector, location, etc.), there are probably as many sets of explanatory
variables as research questions can be formulated. Firms’ innovative performance is the
result of multiple and simultaneous causes converging. However, paraphrasing Robinson
(1962), a model that includes—if possible—all explanatory variables would be as useful as
“a map at the scale of one to one.” The model I will present intends to reduce reality to a
set of key dimensions that account for the main elements highlighted in the literature, not
only in terms of the innovative dynamic but also in terms of the empirical evidence for
unstable environments.
In short, the conceptual framework presented in this chapter hopes to contribute to the
understanding of micro-heterogeneity by following a strategy-based approach. It provides
some key interactions and dimensions that explain different innovative paths. It also aims
to provide a methodological framework to analyze the innovation dynamic at the firm
level, where the firm’s observable characteristics are the result of feedbacks and learning
processes that arise from the interaction between strategy, structure, and capabilities. The
research questions that guide this chapter are related to the understanding of the firms’
strategies and how they change and adjust these. Of course, it also aims to provide
operational definitions for strategy, accumulation, feedback, and firm-level learning
process.
The remainder of this chapter is as follows. In Section 4.1, key methodological
contributions and empirical evidence are presented and analyzed. Section 4.2 contains a
discussion of how strategies, results, and impacts can be translated into methodological
hypotheses, and the general model is presented. Section 4.3 provides some conclusions
regarding the conceptual and methodological approach, which set the basis for the
Chapter 4: Towards a conceptual framework for innovative strategies
- 120 -
empirical testing of the model, which will be performed in Section III. Finally, some
conclusions are provided.
4.1. Literature review: strategies, results, and impacts
According to the conclusions of the previous chapter, three issues need to be considered
when analyzing a firm’s strategy. First, a firm’s strategy is an ex post concept. This means
that although it can be characterized in retrospect, it does not necessarily arise from a
conscious planning and decision process at the firm level. Second, the firm’s strategy can
be observed by looking at its structure and capabilities. At a given point in time and space,
the firm’s structure (resources and investments) accounts for the implementation of the
strategy, while its capabilities are responsible for the performance of both that resource
allocation and possible future strategic adjustments. Of course, present resources and
capabilities are also the result of past strategies. Third, given the complexity of the
innovation process, there are non-observable characteristics, usually summarized as “the
idiosyncrasy of the firm”. The idiosyncrasy of the firm is an emergent result of the
interaction between the firm level and the environment (national, regional, sectoral, etc.)
which affect the strategy, structure, and capabilities, and which lie beyond the scope of any
possible classification (for instance, the individual character of the owner and the culture of
the employees).
From an empirical perspective, the multiplication of information at the firm level has led to
several analyses that aimed to test this interactive process between the firm’s strategy,
structure, capabilities, and performance. Given the objective of this thesis, three types of
these empirical and methodological contributions are worth reviewing: those in which
firms are classified according to their innovative attributes (strategies-structure), dynamic
approximations to innovation results (innovation persistence), and those that aim to test
the impact of innovation on the firm’s competitive position (performance). Each provides
useful insights into the possibilities and limitations of the econometric tools that are most
commonly used to measure the innovation process and to assess its impact, as well as the
underlying assumptions associated with their implementation. By doing this, I will also
present a general overview of the neglected role of learning and accumulation processes at
the firm level.
It is worth noting that what will be discussed in the following sections is a selection of the
most relevant empirical analysis of the innovation process, in line with the objective of this
Chapter 4: Towards a conceptual framework for innovative strategies
- 121 -
thesis. It does not intend to provide a complete literature review of innovation studies at
the firm level but instead aims to discuss selected analyses which, because of their
methodological work or their empirical findings, have provided solid methodological and
empirical bases for measuring the micro-dynamic of innovation and growth and the link
between this and the firm’s strategy.
4.1.1. Empirical evidence on innovative strategies
Since the publication of Freeman’s ideal strategy types, several studies have attempted to
characterize the firm’s innovative dynamic. Based mainly on firm-level surveys, these
analyses have focused on classifying innovative inputs and outputs in order to identify
different innovative patterns, on the assumption that innovation has a positive impact on
performance. These studies provide robust evidence for the existence of heterogeneity,
regardless of the level of development of the firm’s country, its sectoral belonging, size, or
capital ownership.
Arvanitis and Hollenstein (1998) and Hollenstein (2000) analyze the innovative dynamic of
Swiss manufacturing and service firms in order to identify “innovation types” and their
relationship with the firm’s sectoral belonging. Based on data from a survey of the
innovation activities of Swiss private enterprises, the authors performed a factor analysis to
identify key innovative variables and then used the results to run a cluster analysis to
identify different ways of innovating. They found that different combinations of innovation
investments, the economic and productive impact of innovations, and interactions with
other NIS agents lead to two high innovative and three low innovative patterns at the firm
level, with specific characteristics in terms of investments and productive scales. Although
there is greater participation of firms from high-tech sectors in highly innovative clusters,
the authors found no significant association between traditional technological or
productive classifications and the type of innovation. They did, however, find a certain
degree of national specificities. They compared their results with those of Cesaratto and
Magnano (1993) for the case of Italy and found them to be similar in terms of the existence
of each cluster, although biased towards country-specific characteristics: there were more
highly innovative firms in Switzerland versus more scale-intensive (low innovative firms) in
Italy.
Based on the information provided by the national Community Innovation Surveys (CIS),
Leiponen and Drejer (2007) examined intra-industrial heterogeneity for a group of Danish
Chapter 4: Towards a conceptual framework for innovative strategies
- 122 -
and Finnish firms in the manufacturing and service sectors. They used the Statistical
Classification of Economic Activities in the European Community (NACE) at five-digit
desegregation in order to test the relevance of Nelson and Winter’s (1982) technological
regimes and the impact of national specificities. Using factor and cluster analysis, the
authors tested for the existence of a latent correlation between innovative inputs and
outputs (expenditure, information sources, objectives and impact of innovations) and
found four clusters: science-based, market-driven, production-intensive, and supplier
dominated. The clusters share the names of Pavitt’s taxonomies since they present the
same innovative characteristics. However, the authors found that sectoral belonging only
explained half of those belonging to each cluster, regardless of the country. Moreover, the
four clusters can be found in each productive sector, which confirms the existence of intra-
sectoral heterogeneity, and therefore the existence of different innovative behaviors beyond
the technological regime.
Clausen et al. (2011) found similar results and reached the same conclusions for a group of
Norwegian firms, also using factor and cluster analysis. In order to study how different
dimensions of the innovative dynamic are combined, and based on information from three
national CISs, the authors found that firms can be classified according to their specific
investments in innovation, linkages, and capabilities. Like Leiponen and Drejer (2007), they
reach a classification similar to Pavitt’s (1984) taxonomy, only partially determined by the
firm’s sectoral belonging: science-based, R&D-oriented, market-driven, and supplier-based.
With the specific goal of testing the relevance of sectoral classifications—especially Pavitt’s
taxonomy—Srholec and Verspagen (2012) tested for the existence of different innovative
behaviors in a large group of firms located in different European Union (EU) countries
that responded to the CIS. Based on a two-step factor analysis and a cluster test afterwards,
the authors based their classifications on the firm’s capabilities regarding different
“‘ingredients’ of innovative conduct” (research, user, external, and production). They found
five clusters: high-profile, user-driven, externally sourced, opportunistic, and low-profile.
Similar to Freeman’s classification, “high-profile” firms are those with strong capabilities
regarding all four ingredients of innovation. User-driven, externally-sourced, and
opportunistic firms specialize in one innovative area: user satisfaction, the acquisition of
exogenously developed technology, and production and low-R&D niches, respectively. As
with previous studies, these clusters are present in all countries and in all sectors.
Chapter 4: Towards a conceptual framework for innovative strategies
- 123 -
In methodological terms, unlike the other reviewed studies, Srholec and Verspagen (2012)
criticize the lack of coherent articulation between theory and practice when it comes to
identifying strategies. They argue that the innovative dimensions are analyzed as if they
were perfectly interchangeable (not performing R&D could be compensated for through
greater interaction with clients). The authors based their analysis on the idea that
innovation is essentially the combination of different dimensions (investments,
interactions, intended results, impact of innovations) but that each dimension can contain
different “ingredients”—for example, the “investments” dimension might contain R&D
and/or capital goods, while “interactions” might include interactions with S&T institutions
and/or with the supply chain. They then proposed a two-stage analysis in which key
ingredients for each dimension were first identified, and then the weight of each dimension
in the innovative behavior was determined. In this way, different combinations were
allowed while acknowledging the fact that innovation is a multidimensional (multi-
ingredient) process. In this respect, their results provide more robust evidence regarding
the complexity of the innovation process as well as more complex innovative dynamics.
Based on Srholec and Verspagen’s (2012) methodology, Yurtseven and Tandoğan (2012)
applied the same factor-cluster analysis to a set of Turkish firms and reached the same
results. The authors found four groups of firms that achieved innovation results: high-
profile (based on the performance of R&D), external-oriented (based on the acquisition of
capital goods and cooperation), market-oriented, production-intensive innovators, and low-
profile (firms with low investments in innovation). Several years earlier, although based on
a less robust methodology, de Jong and Marsili (2006) reached the same results for a group
of Dutch micro-enterprises: supplier-dominated, specialized-supplier, science-based, and
resource-intensive firms. In both analyses, although a sectoral bias is observed, the
different modes of innovation are present in almost every sector.
Finally, three further studies analyzed the existence of intra-sectoral heterogeneity and the
importance of accounting for micro-behaviors: Fraga et al. (2008) for Portuguese firms and
Frenz and Lambert (2009) and Arundel and Hollanders (2005) using firms from different
OECD countries. By combining the inputs and outputs of the innovative process, they
found different types of innovative clusters, where each cluster is the result of the intensity
of the innovative process (expenditure level) and the decisions regarding how to
incorporate knowledge (make versus buy). In the two first studies, the firm’s expenditure
on R&D and capital goods and its product versus process orientation lead to different
Chapter 4: Towards a conceptual framework for innovative strategies
- 124 -
innovative dynamics, regardless of its sector and national belonging. The results of Arundel
et al., however, show that innovative behaviors (technology developers, modifiers, and
adopters) are associated with the different forms of work organization identified at the
country level, which accounts for one of the links between the NIS and the average firm
behavior.
Aside from the specific name given to each type of innovative behavior, the evidence
presented supports the idea of the existence of heterogeneous situations in terms of the
firm’s innovative dynamic. At the same time, the fact that high-profile and low-investment
innovation behaviors were simultaneously found in countries with different levels of
development contributes to the hypothesis of the possibility of escaping from
environmental determinants. Conversely, the fact that low-profile conduct is present in
dynamic sectors/countries contributes to the hypothesis that a technological regime or
“good” environment are not necessarily enough to trigger high-profile innovative
dynamics.
4.1.1.1. Discussion of the evidence on innovative strategies
From a methodological perspective, the reviewed studies applied the same two-stage
procedure: a factor analysis of several innovative variables followed by a cluster analysis
using the result of the previous stage. A multinomial regression is sometimes also run to
test the robustness of the estimations. Besides Srholec and Verspagen’s (2012) criticisms
regarding the assumption of perfect substitution among the different innovation activities,
the weak sectoral or size controls bias conclusions towards results that are to some extent
tautological: larger high-tech firms are more prone to investing in innovation. Larger firms
will probably have greater incentives to invest in process innovation, given the productive
scale, and they will also have higher probabilities of investing in R&D since they can afford
higher levels of uncertainty without jeopardizing their own survival (Freeman and Soete,
1997). Sectoral belonging will also impact the probability of performing R&D or investing
in capital goods. A firm operating within a dynamic technological regime (Nelson and
Winter, 1982) or in sector where technology is new and the dominant design has not been
imposed yet (Utterback and Abernathy, 1975) will face different incentives than a firm
operating with mature technologies. In other words, the lack of structural control variables
leads to the obvious explanatory factors that shape each strategy without explaining why
some firms do not conform to the expected behavior.
Chapter 4: Towards a conceptual framework for innovative strategies
- 125 -
Another limitation of the analyses is that they focus almost entirely on more standard ways
of incorporating knowledge—R&D and capital goods—to the detriment of other ways of
creating knowledge—such as by improving the firm’s capabilities. The acquisition of a
machine will have a different impact if the firm has capabilities for selecting and adapting it
or if the firm has also carried out training expenditures. Interactions with R&D institutions
will not be the same if the firm has a R&D lab or if it does not, and interactions with
suppliers and customers will trigger different learning process depending on how the
marketing department is organized and its ability to interact both internally and externally.
Of course, the selection of the dimensions used is partly to do with the research questions
and the availability of information, however, when key aspects of innovation are left
aside—mainly those related to the DUI mode of learning—a large part of the existing
heterogeneity cannot be observed. This leads over and over again to the same “make
versus buy” conclusions: some firms base their innovation process on the knowledge
production through R&D (make) while others base it on external acquisition by
incorporating capital goods (buy). As we shall see, it is the combination of both activities
that makes the difference in terms of innovation results and impacts.
The simultaneous inclusion of input and output dimensions as classifying criteria is another
aspect that limits the analysis. In methodological terms, including inputs and outputs on the
same side of the equation reinforces the correlation of variables, which leads, once again, to
tautological results: a firm that achieved process innovations will probably have invested in
the corresponding innovation activities (capital goods), and a firm that performs R&D has
higher probabilities of product innovation than a firm which does not. In theoretical terms,
this type of analysis restricts the possibility of capturing complementary learning processes
that arise from the firm’s innovative dynamic, which feed back into the general level of
capabilities: problem-solving quests could trigger process innovations but also improve
product characteristics and allow the firm to increase its market share. A process
innovation could alter product characteristics and eventually lead to a product innovation.
Consequently, the unidirectional association between types of inputs and types of outputs
works against one of the main features of innovation: the more than proportional impact it
has on the firm’s competence building.
In this respect, the distinction between strategy, structure, results, and impacts seems to
allow firm behavior to be better understood. To some extent, the names given to each
cluster in the reviewed literature resemble this study’s concept of innovative structure, in
Chapter 4: Towards a conceptual framework for innovative strategies
- 126 -
the sense that they refer to the set of observable elements regarding an innovation-based
way of facing the selection process (usually referred as inputs of the innovation process:
investments, sources of information, linkages, R&D personnel). However, structure,
results, and impact are mixed up and this works against the understanding of the complex
nature of the innovation process and how it impacts the firm’s performance (productivity,
sales, exports) whether the firm has innovated or not (achieved results).
Product or process innovations, patents, and organizational changes are results of the
innovation process, while the share of sales of innovative products, productivity
improvements derived from new processes, or new commercialization techniques are the
impacts. In the reviewed analyses, when these different dimensions are taken together
(structure, capabilities, results, and impacts), the innovative process is understood as being
linear and unidirectional (if the firm invests in R&D, then the impact is measured in
patents; if the firm invests in capital goods, it is measured in process innovations), which
neglects the interactive and multi-casual dynamics of innovation (Kline and Rosenberg,
1989). By insisting on the one-way nature of this relationship, the studies in question rule
out the possibility of one innovation generating additionality effects in other competitive
dimensions of the firm (a product innovation that leads to a new commercialization
channel, or a process innovation which allows a marginal product innovation). Of course,
these analyses also neglects the fact that marginal innovations could have been triggered by
problem-solving routines that are not necessarily associated with a specific innovative
effort but are allowed by the firm’s capabilities (Nelson and Winter, 1982).
The last limitation of the reviewed literature worth mentioning is the static nature of the
analysis. If the firm has invested in the search for a new product and successfully
introduced it to the market, it will probably invest in marketing or process innovations in
the next period to maximize the impact of the original innovation. If the firm did not
succeed with the innovation, it will probably change its strategy and this will lead to
changes in the structure. Static analyses provide valuable information regarding the
situation of a particular innovative structure in a particular time and space, but they fell
short of providing elements to predict how firms will react to the different incentives, or, if
given the same incentives, whether the firm will continue with the same innovative strategy
or not. In this respect, studies about the persistence of innovation are a good
approximation to the innovation process from a dynamic perspective as long as they are
based on the idea that past actions affect present results.
Chapter 4: Towards a conceptual framework for innovative strategies
- 127 -
4.1.2. Innovation results: persistence of innovation
The review of the literature that aims to test the persistence of innovation (persistence of
positive results) shows that there are two types of analyses, which coincide with the
availability of information at the time they were performed. On the one hand, there are
persistence studies based on the relationship between past and present innovations but that
take into account the firm’s innovative structure (input variables) and type of innovation
results (output variables). These studies are based on the merging of innovation surveys
and they are usually referred to as input-output studies. On the other hand, there are the
output studies, based on patents and other administrative records. Although not all
innovations are patented and not all patents are transformed into innovations, the
advantage of these studies is that they observe trajectories in different productive sectors in
different countries, allowing for cross-sectoral and cross-national comparisons.
Among the first type of studies is the one performed by Raymond et al. (2010), for a set of
Dutch firms and with information provided by three rounds of innovation surveys. The
authors found that persistence exists among firms from high and medium-high
technological intensity sectors, where—besides size—R&D investments and access to
public subsidies have a significant positive effect on the probability of reaching
innovations. In the other sectors (medium-low and low technological intensity), the
hypothesis of persistence is not verified. Moreover, when the initial condition (the firm’s
idiosyncratic characteristics) is not properly controlled, the authors note the existence of
“spurious persistence,” related to the presence of gaps in the characterization of the firm
when designing the model. In other words, when the initial condition and other
unobservable factors are not properly accounted for, the relationship between past and
present innovations may actually be a manifestation of the statistical correlation between
the firm’s idiosyncratic characteristics and its success as a firm that achieved innovation
results. As a consequence, the authors suggest that in addition to structural controls (size,
sector, capital ownership), for any empirical testing of persistence or any other innovative
dimension of the firm’s dynamic, the initial condition and the unobservable effects should
be controlled.
Using panel data about German firms, Peters (2009) establishes the causality between past
and present actions, using innovation investments as result proxies and including Raymond
et al.’s (2010) recommendations for unobservable effects. Similarly to the results obtained
by these authors, while controlling for the effects of the initial condition, Peters
Chapter 4: Towards a conceptual framework for innovative strategies
- 128 -
corroborates persistence and also finds that size and access to subsidies are relevant
variables to explaining the continuity of investments. She also notes a strong correlation
with other firm attributes: its capabilities, the capital ownership of the firm, and its degree
of exposure to international competition.
In another study based on innovation surveys, this time for the case of Luxembourgish
firms, Le Bas et al. (2011) analyze persistence but differentiate between three types of
innovation profile: firms that innovated on product or process in the two periods under
study, those who innovated only once, and those who did not achieve innovation results at
all. Using this distinction, the authors test the impact of organizational innovations in firms
in which persistence has been verified (the first group of firms) and in which it has not (the
second and the third groups). The evidence shows that organizational innovations are a key
determinant not only of persistence, but of innovations generally speaking. In this sense,
the authors observe significant and positive impacts that derive from this type of
innovation in both firms with continuous and sporadic innovation results. As in the other
studies reviewed here, there is also a strong association between persistence and the firm’s
size and R&D investments.
In the abovementioned study by Clausen et al. (2011), which aimed to study how different
dimensions of the innovative structure are combined inside the firm, the authors analyze a
group of Norwegian enterprises to test the relationship between innovative structures and
persistence. Using a probit dynamic model, they corroborate persistence in the case of
science-based and market-oriented firms, but not in the case of firms with sporadic
innovation investments (the ad-hoc group), or with supplier-based ones. They also confirm
that the elements of the innovative process are combined differently to determine different
structures. As such, these elements (investments, linkages, and capabilities) should not be
studied as a single one-dimensional variable but as part of more systemic behavior where
different investments are combined with different capabilities, and both investments and
capabilities are combined with different interactions with the NIS.
The work of Malerba et al. (1997), Cefis and Orsenigo (2001) and Geroski (1997) is among
the second type of persistence analysis, based on output data. In all three cases, the
explanatory power of past innovations—approximated through patents—is low and
focused on some productive sectors. Malerba et al., however, note that although the
percentage of firms persistently innovating is reduced, these firms account for most of the
granted patents. Cefis and Orsenigo, meanwhile, point out that there is a strong tendency
Chapter 4: Towards a conceptual framework for innovative strategies
- 129 -
towards the perpetuation of the status of firms with positive innovation results as well as
the status of non-innovator firm, although the strength of this tendency decreases over
time. Finally, Geroski’s results on U.K. firms that obtained patents in the United States
agreed with those obtained by Malerba et al., in that he found only a few companies which
innovate persistently.
4.1.2.1. Some gray areas in studies about persistence
Persistence studies are one of the few dynamic approaches to innovation at the firm level
where the impact of feedbacks is acknowledged and estimated. The main strength of these
studies lies in how theoretical propositions are matched with econometric tools. Firstly, the
use of time series and panel data econometric models allows serial correlation to be
introduced among variables, which coincides with the idea that firms have path
dependence trajectories. Secondly, the inclusion of unobservable characteristics controls
for firm-specific features, which is in line with the theoretical argument regarding the
impact of the idiosyncrasy of the firm. Thirdly, despite the unavoidable endogeneity
problem that could arise from the bidirectional relationship between innovation outputs
and inputs, these studies offer a more complex analysis of the firm and the dynamic nature
of the process that links these two concepts.
In terms of their findings, all the studies reviewed show a positive relationship between
investments in innovation activities, firm size, sectoral belonging, capabilities, and recurring
innovation. All the same, the evidence is inconclusive regarding the relationship between
past and present innovations. The diversity of results shows that despite the significant
positive impact of the estimated coefficients, there are firms that achieve positive results
year after year and firms that do not, regardless how frequently or intensely they invest in
innovation. There are also firms that achieved innovations at the beginning of the time
series but did not do so again (and vice versa). Most of the time, persistence is restricted
only to certain groups of firms.
Theoretically speaking, the lack of absolute confirmation of persistence is not surprising.
On the one hand, this is because of the uncertainty attached to the innovative process; on
the other hand, because the selection process is not exclusively determined by market
institutions. As long as the firm moves forward with its innovative strategy, the increase in
the information decisions are based on, learning processes, and the evolution of the
environment will lead the firm to adjust its strategy (Chapter 3.3). Within this context, the
Chapter 4: Towards a conceptual framework for innovative strategies
- 130 -
innovative process will trigger additional learning and accumulation processes that will
enhance the firm’s capabilities and increase its resources. As a result, the firm will more
likely to achieve positive results in the future. However, the characteristics of that
persistence (level of probability, type of innovation, and magnitude of the impacts) will be
explained by the specificities of firm behavior (innovation activities, capabilities, resources)
and how it deals with the environment (at the national, meso, and regional level). It follows
that innovation persistence is the result of sustained innovative behavior and can be
explained by the impact of path dependence (Antonelli, 1997), cycles of accumulation
(Nelson and Winter, 1982), and the increase in market power (Geroski et al., 1997).
However, depending on the firm’s capability for reacting to signals from the environment
and from its own learning processes, changes in firm behavior can take place from one
moment to the next: innovation projects can be abandoned, the firm can change the
sought-after innovation, or a past innovation may no longer be suitable for the
environment. In all these cases, the expected feedback process may never happen. The
same reasoning applies to a change in the environmental situation. Faced with changes in
the environment, the firm may continue or change its innovative behavior.
In this respect, although the reviewed literature argues that path dependence will limit the
range of the firm’s possible responses, and this could work against innovation persistence,
it neglects the possibility of a change in the firm’s innovative behavior. If the firm was
exploiting an innovation that was no longer useful for the new environment, then past
results and past learning and accumulation processes could delay the achievement of future
innovations. In this case, the firm would have to adapt its innovative behavior, undertake
new investments, and acquire new capabilities before expecting future results. Conversely,
if the firm was not innovative, and the new environment created incentives to innovate,
then present innovations would arise from path dependence without innovations, and
persistence would not be confirmed. None of these possibilities are contemplated in the
persistence literature. Moreover, the possibility of a change in the firm’s innovative
behavior is ruled out of the research design.
Another element excluded from persistence studies is that past innovations can be a chance
event that was triggered by a specific problem-solving process (Nelson and Winter, 1982)
or the search for short-term benefits following opportunistic behavior (Freeman, 1974). In
these cases, although learning and accumulation processes may arise, arguments regarding
the impact of past innovations on present ones are weak, to say the least. This is especially
Chapter 4: Towards a conceptual framework for innovative strategies
- 131 -
true in the case of periods of changes in the environment, where levels of uncertainty
create incentives to look for short-term improvements that are not necessarily based on
innovation (Fajnzylber, 1989; Kosacoff and Ramos, 2006; Ocampo, 2012). Consequently, if
persistence is empirically confirmed, there would be good reason to believe that it is a
spurious type of persistence in the sense of Raymond et al. (2010), caused by the variables
that have been omitted from the model.
Methodologically speaking, the lack of generalized confirmation about persistence opens
up new questions regarding how a dynamic, interactive, and evolving process can be
formalized into a mathematical model. Three main limitations can be observed: the timing
between inputs and outputs, the use of innovation results as proxies of performance, and
the narrow consideration of environmental impact.
One of the unsolved problems of persistence studies has to do with determining the lag
between the dependent and the explanatory variables (when innovations become past
events and when innovation investments are supposed to impact on results). In practice,
the estimated lag seems to be explained more by the availability of information than by a
theoretical framework for the time window considered in the analysis. At the same time,
the return period of R&D or machinery expenditures is not necessarily equal to the time
that it takes to reinvest profits or to the enhanced capabilities for generating new
innovations. Moreover, the time that R&D investments take to impact innovations is
probably different from the time that it takes new machinery to impact the productive
process. Evidently, an average delay is better than no delay at all, and only the increase in
the quantity and quality of available information will allow more complex approaches to
the relationship between past actions and present results. However, the limitations given by
the arbitrary selection of lags cannot be neglected and should be considered when
interpreting results.
The second limitation of persistence studies arises from the use of innovations as a proxy
for success (better performance). The presented analyses argue that feedbacks generated
from an innovation breed profitability and this enables future innovations. However,
although the correlation between innovation investments, innovations, and capabilities is
widely supported, international comparisons draw attention to the use of innovation as a
synonym for success. Indeed, when looking at the share of firms with innovation results in
the world, there is no clear correlation between this share and performance levels
(measured as productivity, exports, or labor). Argentina, Brazil, and Uruguay, for example,
Chapter 4: Towards a conceptual framework for innovative strategies
- 132 -
have a greater share of firms with innovation results than developed countries like
Germany or France59 but they have lower levels of national income. In this sense, although
innovation is the successful introduction of a product or process to the market, the
implementation of a new product does not automatically lead to improving the firm’s
performance, but nor does its absence prevent this. On the contrary, it is precisely what
happens between the input and output—learning processes—which allows the impact of
the innovative dynamics to be understood.
Finally, as with the limits of studies about innovative behaviors, another aspect still under
discussion is the possibility of a change in firm behavior. When analyzing the phenomenon
of persistence, models neglect the possibility of entering into the “club of innovator firms.”
Furthermore, although micro-heterogeneity is considered in the theoretical framework, it is
not included in the empirical testing. Worse still, the existence of firms that change their
strategy works against the confirmation of the hypotheses. Although some of the reviewed
analyses consider changes in the environment that vary over time, it is also assumed that
the impact is the same for all cases (time-variant case-invariant variables). This implies that
the incentives derived from the environment impact equally in all firms, a widely criticized
assumption to the standard neoclassical economics. In other words, the dynamic aspect of
the persistence literature is just the inclusion of past events into the model, while the
learning and accumulation processes that emerged from those events and different impacts
from changes in the environment are neglected.
To conclude, studying persistence within a context of environmental change leads to an
expectation that firms will modify their behavior and that this will affect the persistence of
innovation. Some innovative strategies can trigger the expected virtuous process of
accumulation that leads to persistence, but others can lead to results with low impact on
future innovations. As a consequence, certain innovative strategies will lead to the firm
successfully innovating, but only some firms to innovate persistently. Within unstable
environments, persistence should be the manifestation of specific innovative strategies
derived from resources and capabilities that allow the firm to deal with the environment
rather than a simple serial correlation between past and present results. From a theoretical
59 Another interpretation of the lack of correlation between the rate of firms with positive innovation results and the level of development has to do with the quality of the data gathered by the innovation surveys in developing countries. Questions related to innovation outputs are subject to the respondent’s understanding and his knowledge of the international technological frontier. As a result, even when the questionnaires follow the Oslo Manual recommendations, the results of cross-country comparisons ought to be interpreted with caution.
Chapter 4: Towards a conceptual framework for innovative strategies
- 133 -
angle, this means that the feedback and accumulation processes that the literature attributes
to the persistence of innovation (persistence of results) are in fact the result of the
persistence of certain innovative behaviors.
4.1.3. Innovative impacts: productivity and growth
Among economic studies, there is a long tradition of studying the impact of technological
change on the firm’s performance. Causal relationships between performance and the
improvements of techniques or new product development can be traced back to the
writings of Adam Smith (1776) and Karl Marx (1858) and, of course, Schumpeter (1912).
Within modern studies, several approaches have tried to establish how economic growth
can be explained in terms of technological improvements in the production function and
the existence of endogenous and exogenous factors that shape the relationship.60 Within
the Schumpeterian tradition, Freeman’s (1974) innovative strategies, Perez and Soete’s
(1988) windows of opportunity, Dosi’s (1991) model of growth, and other contributions
from economic history literature are just some examples of the vast literature on the impact
of technological change on growth.
At the firm level, the study of the impact of innovation on performance evolved as a result
of developments in the available information. During the 1970s and 1980s, most studies
were based on aggregate data on R&D expenditure and patents, in some cases crossed with
sectoral or geographical specificities.61 During the 1990s, the first innovation studies at the
firm level appeared, based on national innovation surveys.62 This last group of studies
confirmed the findings of the previous ones and proved not only the existence of a positive
relationship between innovative structure and different measures of performance (sales,
productivity, labor, exports) but also the higher mortality rate among those firms not
engaged in innovation. However, unlike the previous generation of studies, the analysis of
information at the firm level shed light on the existence of multiple explanatory factors for
innovation results besides R&D and investments in machinery. Moreover, they proved that
innovation results are more than patents and new products, in that organizational practices
and marginal improvements had also led to different performance levels among firms.
60 See Solow (1956), Meade (1952), Nurkse (1961), and Romer (1990), among others. 61 Utterback and Abernathy (1975), Mowery and Rosenberg (1982), Pavitt (1984), Rossegger (1987), Dosi
(1988), and Kline and Rosenberg (1989), among others. 62 Mairesse and Sassenou (1991), Nelson (1993), Alburquerque (1999), Bisang et al. (2002), Godinho et al. (2004), Malerba (2004), among others.
Chapter 4: Towards a conceptual framework for innovative strategies
- 134 -
During the 2000s, the increase in the availability of information from innovation surveys,
the improvement of econometric techniques, and the standardization of software tools led
to the spread of a new set of studies about the impact of innovation on performance.63
Besides the possibility of comparing developed and developing countries, these studies
provided a deeper understanding of the impact of innovation and the variables that best
explained this relationship (mainly investments, capabilities, linkages, and funding). In turn,
they provided solid proofs regarding the existence of heterogeneity at the firm level and the
partial impact of the firm’s sector, size, and location.
After more than 50 years of innovation studies, there is no doubt that innovation positively
impacts performance and that a significant number of variables affect this relationship.
However, the intensity of the interaction between the multiple innovative inputs (the
different innovative structures), the firm’s characteristics, and environmental determinants
are still the subject of research and discussion among innovation analysts. In this sense,
discussing the state-of-the-art in the matter of innovation and productivity is not an easy
task, given the various empirical and methodological contributions now available. To
approach this matter, I will refer to the two cornerstones that have guided the majority of
these analyses within evolutionary studies: Nelson and Winter’s (1982) contribution
regarding the sequence of events that lead to extraordinary profits and the model of
Crépon’s et al. (1998) to empirically test this sequence.
4.1.3.1. Nelson and Winter’s relationship between innovation and performance
The first cornerstone for the analysis of the impact of innovation on performance is
Nelson and Winter’s (1982) classic book An Evolutionary Theory of Economic Change. In it, the
authors studied the evolution of market structures—in terms of sales concentration—in
order to analyze how innovation impacted the firm’s performance. From a panel of U.S.
firms in different industries, the authors conclude that there is a strong trend towards
increasing the market share among firms with a more intense innovative dynamic
(capabilities and investments), but that it does not happen at the same pace or with equal
intensity in all markets, given the different technological opportunities. Their study presents
similar conclusions to those of Mansfield (1962), who found that concentration due to
technological change becomes more pronounced along the technological path but with a
decreasing rate. Besides the impact of investments and capabilities, Nelson and Winter
63 Kemp et al. (2003), Chudnovsky et al. (2004), Benavente (2004), De Negri et al. (2005), Lugones et al. (2005), Hall et al. (2006), Goedhuys (2007), Jensen et al. (2007), Iacovone and Crespi (2010), Ito and Lechevalier (2010), among others.
Chapter 4: Towards a conceptual framework for innovative strategies
- 135 -
noted that firm size and sector were determining factors and were strongly correlated with
the increase in market share. In this sense, they confirm the hypothesis of Phillips (1971),
who had already stated that the phenomenon of concentration was greater where the
technological opportunities were lower.
To date, Nelson and Winter’s findings in terms of the impact of productive activity and the
higher probabilities of innovating among larger firms have been confirmed by several
studies, some based on the sectoral impact of technological change (Breschi and Malerba,
1997; Malerba, 2004, Nelson and Malerba, 2007 among others), and others on the impact
of the firm’s scale, the level of internationalization, and the specific characteristics of SMEs
(Freeman and Soete, 1997; Lall, 2001; Lall and Dunnewijk, 2008; to name a few). At the
same time, the positive relationship between the innovation process and higher profit levels
has not been refuted up to now, although additional considerations regarding how the
selection process leads to differential performances have to be made.
In this respect, Bottazzi et al. (2010) provide evidence regarding how Nelson and Winter’s
competition process operates. The authors analyze the relationship between efficiency,
competitiveness, and profitability within the process of the firm’s market selection and
survival. Using data on Italian and French manufacturing firms, the authors find that the
market selection process is not as “efficient” as Schumpeter sustained as long as firms with
heterogeneous levels of productivity and profitability remain in the market. That is, the
slowdown of the concentration process observed by Nelson and Winter does not
necessarily lead to the disappearance of less competitive firms but to the stabilization of
heterogeneous situations once the first shock caused by the incumbents (or the radical
innovation) has shaken up the status quo.
In terms of innovative dynamics, and also following Nelson and Winter’s evolutionary
approach, Dosi et al. (2010) review the literature on firm performance and the persistence
of heterogeneity and find three factors explaining the lack of convergence: the quality of
inputs (workforce, management, and capital), innovation investments, and firms’
idiosyncratic characteristics. Particular combinations of these factors enable some firms are
able to survive with lower productivity levels. The authors sustain that, given the firm’s
understanding of the environment and given its resources and capabilities, there is a self-
selection process where firms choose the market they want to compete in and the type of
innovative trajectory they want to follow. Afterwards, interactions with the rest of the
system, technological and commercial opportunities and constraints, and the firm’s level of
Chapter 4: Towards a conceptual framework for innovative strategies
- 136 -
capabilities will determine their survival. In this sense, innovation impacts performance but
at different intensities and paces from firm to firm, depending not only on the firm’s
innovative investments, size, and sector, but also on its capabilities for dealing with the self-
selection process (to make the most adequate choice).
From this literature, it can be concluded that the relationship between innovation and
performance is a process that starts with decisions about investments in innovation (self-
selection in Dosi et al.’s (2010) terms, the innovative strategy in this study), is followed by
the implementation of those decisions (the structure), and ends with the achieving of extra-
profits (innovation results and impact on performance). This process is mediated by the
firm’s capabilities, the competition dynamic of the market, and the decisions the firm takes
about strategy adjustment. However, although higher profits are expected from an
innovative project, achieving positive results does not guarantee better performance, nor
does the lack of innovation investments leads to the firm’s disappearance from the market.
4.1.3.2. Crépon, Duguet, and Mairesse’s model of results and productivity
The second cornerstone of the studies about innovation and performance is the work of
Crépon, Duguet, and Mairesse (1998). Their work set the basis for the bulk of the
subsequent analyses of the relationship between innovation and productivity, to the point
that the methodology they used is now known as the CDM model.64 In this sense, similar
analyses of the impact of innovation on performance can be found, with similar results for
both developed and developing countries: innovation results impact productivity and this
can be observed through the innovation process.
With the objective of characterizing the triple interaction between innovation inputs,
outputs, and impact (structure, results, and performance, in this study), Crépon et al. (1998)
constructed a model to test the sequence of events that cause innovation to improve the
firm’s productivity. The authors claim that it is not R&D that impacts productivity but the
results of R&D activities, namely, innovations. Using data about French manufacturing
firms and the implementation of instrumented variables, Crépon et al. constructed a model
to simultaneously test the impact of innovation investments (R&D) on innovation outputs
(patents), and innovation outputs on productivity (sales and total factor productivity), while
controlling for firm-specific and structural variables (size, sector, technological
opportunities, and demand). Besides corroborating the positive relationship in both cases,
64 Chudnovsky et al. (2004), Benavente and Contreras (2006), Crespi et al. (2007), Castellacci (2010), Yoguel et al. (2011), and Huergo and Moreno (2011), among others.
Chapter 4: Towards a conceptual framework for innovative strategies
- 137 -
the authors’ main contribution is the empirical confirmation of the bidirectional
relationship between innovation investments and results, and between results and impacts.
They therefore confirm the existence of endogeneity between the main innovation
variables.
Methodologically speaking, Crépon et al.’s (1998) model for controlling endogeneity
between innovation results and the firm’s performance is based on the use of simultaneous
equations and instrumented variables. The theoretical interpretation of their findings, as
discussed above, centers on processes of feedback and accumulation and the importance of
the firm’s path dependence and capabilities when it comes to estimating the marginal
impact of an additional unit of investment on innovation (Yoguel et al., 2011). In other
words, innovation is an interactive and evolving process, and in this sense, there are some
problems with CDM models in terms of the articulation between theory and practice.
One limitation of the CDM models is that they do not account for the impact of the co-
evolution of innovative dimensions, to the extent that the innovation inputs they
contemplate are basically R&D investments. For instance, Percival and Cozzarin (2008)
study the impact of the innovation process on performance for a group of Canadian firms,
using information provided by the national innovation survey (similar to the CIS). The
authors found that the better articulated the innovative process, the greater its impact in
terms of the innovations reached. Firms in which innovative inputs arise from a
combination of different complementary activities (innovation investment levels, the
objective of the sought-after innovations, interaction with other organizations, etc.) present
higher levels of impacts in terms of the novelty and scope of the innovations achieved.
Another limitation is the static nature of the model. In this regard, based on the World
Bank survey about the investment climate for Brazilian firms, Goedhuys (2007) finds that
the positive impact of investments in R&D can only be observed over long periods, while
in the short-term, investments in training and machinery have the greatest impact on
performance. From a theoretical perspective this means that innovation is a process that is
not necessarily equivalent to the calendar year or the reference period of the surveys, and
the time between the investment and the impact could differ depending not only on the
type of innovation investments but also the stage of the innovative process at the firm
level.
Chapter 4: Towards a conceptual framework for innovative strategies
- 138 -
In conclusion, it can be stated that the main contribution of the CMD model is its ability to
account for the sequence of the innovation process, the results of this, and its impact on
performance. However, the model’s static approximation to innovation impact draws
attention to how the results of static analysis are interpreted. Although they are useful
inputs for descriptive analysis, they could lead to misleading prescriptive conclusions in that
they cannot predict how firms will react to the innovative and productive process, still less
how a set of innovative behaviors will evolve. As we shall see, allowing for changes in firm
behavior is a key aspect to understanding how firms learn to innovate, grow, and improve
their productivity and how the heterogeneity of behaviors changes over time.
4.1.4. Environmental determinants: structural variables
Specific characteristics of the environment in which the firm operates have a positive
impact on the firm’s innovative performance in all the analyses reviewed. Thus, empirical
testing of the innovative dynamic at the firm level demands the inclusion of a set of
variables capable of accounting for the meso- and macroeconomic determinants, which
were referred to above as “structural determinants.” Case-based, sectoral, and national
studies have significantly contributed to the understanding of the innovative dynamic, and
also to the understanding of micro-heterogeneity (e.g. Borello et al., 2006; Cassiolato and
Lastre, 2005; Edquist and Lundvall, 1993; Malerba, 2004;, Dutrenit 2009; Nelson and
Malerba, 2010; Porta et al., 2011; among many others). They have proved solid arguments
regarding the importance of the national situation, and the firm’s sectoral belonging, the
capital ownership and the size.
Structural variables—those highlighted by the NIS approach—account for the national
characteristics and the impact of history. At the macro level, this implies that the
interpretation of results in one country could be different from the interpretation in
another. It also highlights the importance of comparing similar macroeconomic situations.
The impact of innovation during a period of strong recession will be different from a
period of growth. Likewise, the meaning of persistence of innovation will not be the same
between two moments with the same exchange rate than between periods when a
devaluation of the domestic currency took place.
In this sense, variables such as the inflation rate, GDP evolution, unemployment levels, and
the exchange rate are all factors that influence the firm’s decision to continue with an
innovative strategy or not. Unlike the structural historical factors influencing innovation
Chapter 4: Towards a conceptual framework for innovative strategies
- 139 -
that were discussed with regard to the NIS approach (the labor market, the education
system, the competition regulation; see Chapter 1.2.2.3), the macroeconomic variables
change within shorter periods of time—even shorter in the case of unstable
environments—and generate specific (dis)incentives to invest (Katz, 2000; Kosacoff and
Ramos, 2006; Katz and Bernat, 2011; see Chapter 1.3.1). Therefore, special attention
should be paid to how to control for these macroeconomic incentives.
At the meso level, the firm’s sectoral belonging and size are probably the most important
factors that influence the firm’s innovative dynamic. The former is usually approximated by
the inclusion of industrial classifications, in order to control for technological opportunities
and the general evolution of the domestic and external demand. In turn, the firm’s size is
controlled by including labor or sales variables and accounts for the scale determinants
(Crépon et al., 1998; Yoguel et al., 2011).
Firm ownership is another dimension that accounts for part of the innovative dynamic.
The most obvious characteristic is that the strategy of a foreign capital–owned firm is
beyond the productive and innovative dynamic of the local subsidiary. How foreign capitals
are regulated, which is the advantage that the firm wants to exploit (natural resources,
access to regional markets, cheap labor), and the type of national environment of the
subsidiary’s headquarters are just some of the factors that will shape the innovative
behavior of the local establishment captured by national analysis (Bell and Albu, 1999).
Consequently, studying innovative behavior at the firm level implies acknowledging the
existence of national and sectoral specificities that shape micro-behavior. The challenge for
this type of studies is to try to go beyond these structural determinants. That is precisely
the main objective of this thesis: to observe how and why firm trajectories differ from that
predicted by the NIS approach. In terms of the present analysis, the inclusion of sectoral
control variables is a way of controlling not only the technological and scale characteristics
of the sector but also how it has evolved in a particular environment; the different public
policies that have fostered or limited its growth; how the science, technology, and
educational system is articulated at this meso level; the natural competitive advantages, etc.
Methodologically speaking, this means that the lack of structural controls could bias the
results and lead to the sectoral or size biases identified in most of the reviewed analysis.
Not to include these controls is to assume that micro-heterogeneity will be the same across
sectors or sizes.
Chapter 4: Towards a conceptual framework for innovative strategies
- 140 -
4.2. Conceptual approach and discussion of the hypotheses
The main motivation of this thesis is to test the existence of different innovative strategies,
some of them better to cope with the unstable environment, through which firms managed
to successfully innovate and grow. As discussed thus far, these different innovative
strategies are the result of the firm’s past decisions regarding how to face the selection
process, decisions which are both cause and consequence of the firm’s structure and
capabilities, and the history and characteristics of the environment. As a result of this
motivation and the proposed strategy-based approach to firm behavior, a model aimed at
measuring the relationship between the innovative dynamic, the firm’s innovation results,
and its economic performance will be constructed.
The conceptual approach adopted in this thesis and derived from the strategy-based
approach is presented in Figures 4.1.a and 4.1.b and can be summarized as follows. Firms’
decisions regarding how to face the selection process constitute the firm’s competitive
strategy and impact its general structure and capabilities. When the firm decides to compete
through innovation, it is pursuing an innovation-based strategy. Therefore, the innovative
strategy is part of the firm’s wider competitive strategy (which also includes the decisions
regarding the productive process, the commercial organization, the financial dimension,
etc.). Based on its innovative strategy, the firm will allocate resources to the creation,
absorption, combination, and application of knowledge in order to innovate, which can be
observed by looking at the innovative structure. Of course, the firm may succeed or fail in
achieving an innovation. However, and regardless of the results, additional feedbacks65 and
learning processes will be triggered and the firm’s capabilities will be enhanced. It follows
that those firms with a sustained innovative strategy will have higher capabilities and will be
better prepared to deal with the environment than a firm with a non-innovative strategy
(Figure 4.1.a versus figure 4.1.b). Firms with an innovative strategy should be in a better
position to successfully adapt their competitive strategy to eventual changes in external
conditions or to avoid environmental constraints in order to maintain or improve their
performances. These better performances will feed back into the strategic process and—
together with the characteristics of the environment and the resulting level of capabilities
and resources—will define how the firm adjusts its strategic behavior regarding how to face
the selection process.
65 In order to simplify the graphic representation, only the additional feedbacks that arise from the innovation process were included (the arrows). Feedbacks from the productive process and from the environment were omitted from the scheme.
Chapter 4: Towards a conceptual framework for innovative strategies
- 141 -
Figure 4.1.a: The strategy-based approach—The innovative strategy
Figure 4.1.b: The strategy-based approach—The non-innovative strategy
4.2.1. Innovative strategies and innovative structures
While the reviewed literature assumes—due to data restrictions or for theoretical reasons—
that firms always choose an innovative strategy—-, this analysis will start one step earlier
and will analyze the differential impact of engaging in an innovative strategy or not. This
means that all the different innovative behaviors presented in the literature review in 4.1.1
(market-oriented, science-based, user-driven, high-profile, etc.) are included in what will
here be called the “innovative structure.” They represent different innovative structures
associated with the specific objectives of the innovative strategy and include the innovative
projects the firm is carrying out. In this sense, the approach used in this study allows the
firm to pursue a product innovation in one period and a process innovation in the
following one, an aspect that coincides with the expected adaptation of the firm’s strategy
based on its past results. Of course, the firm can also decide not to continue with the
innovative strategy at all.
At every point in time, the firm faces three options: to start an innovative strategy, to
continue with an existing one, or to stop doing so. Only when the firm has decided to
pursue an innovation-based strategy (starting or continuing to do so), can we look at the
Chapter 4: Towards a conceptual framework for innovative strategies
- 142 -
specific innovation projects it is carrying out, that is, the innovative structure. Therefore,
the definition of innovative strategy used here allows the firm to pursue a dynamic and
evolving strategy while accepting the possibility that the firm might face competition
without innovating. In practical terms, the firm’s strategy will be observed by looking at the
innovative structure, on the assumption that it is the result of past decisions,
implementations, and adjustments of the strategy.
According to what was discussed in Chapter 3.2, the innovative structure is the set of
resources allocated to the quest for innovation, which can be grouped into three categories:
expenditures on innovation (innovation investments), innovative capabilities, and the
remaining innovative resources. Since the firm’s strategy is a continuum of strategic
decisions, implementation, and adjustments, when one looks at firms’ strategic choices in
dynamic terms, firms can change strategies, and this can be observed by looking at changes
in the innovative structure. This means that within a given period of time, the structure is
the result of past strategic choices and will influence on future ones.
According to the literature reviewed in this chapter, the structure of the innovative strategy
can be approached by means of investments in innovation, the characteristics of the human
resources, and the availability of resources to invest. With regard to capabilities, the
innovative structure will be impacted not only by those specific human resources allocated
to the project but also by all individual skills (this is the case, for instance, for Nelson and
Winter’s problem-solving innovation). To narrow down a firm’s capabilities to its R&D
human resources would work against the idea that a firm is more than the sum of its
employees and deny the importance of interactions within the firm. Moreover, it would
imply analyzing the innovative project and not the innovative strategy.
As in any other investment project, expenditures on innovation are the key variable to
understanding the innovative strategy, in that they account for the firm’s commitment to
the quest for innovations (that is why expenditures on innovation are also referred as
“innovation efforts”). The literature refers to this variable as the main input of the
innovative process and measures it in terms of R&D and machinery expenditures. Unlike
in the literature, a wider definition of innovation activities will be used here, which includes
alternatives ways of creating, incorporating, and applying knowledge: engineering and
industrial design, training, consulting, technology transfer, etc.66 In this way, this study’s
approach is consistent with Nelson’s and Lundvall’s ideas regarding the translation of
66 These variables will be further discussed and described in detail in Chapter 5.
Chapter 4: Towards a conceptual framework for innovative strategies
- 143 -
knowledge into profits (problem-solving routines and DUI and STI modes of learning).
This variable—innovation expenditures—will be used as the first classification criteria:
firms with expenditure on innovation are performing an innovative strategy; those without
expenditure are not. If the firm continuously allocates resources to innovation, regardless
of the results, then it is performing a continuous innovative strategy. It is important to keep
in mind that “innovative strategy” refers to the fact the firm is investing in innovation (an
input-based definition). Achieving innovations (results/outputs) is a possible outcome of
this strategy which is not related to how sustained the firm’s investment behavior is.
Both the theoretical analysis and the empirical evidence sustain that expenditures on
innovation should be higher and sustained over time (continuous) in those firms with a
more complex strategy, in the sense of the knowledge that is being generated, applied, and
appropriated. The combination of STI and DIU modes, in Lundvall’s (2009) terms; R&D
and engineering investments in Freeman’s (1974) terms; and the existence of routines to
innovate, in Nelson and Winter’s (1982) terms are just some of the possible explanations
that support the validity of the assumption about the positive relationship between the
continuity of expenditure and the complexity of the strategy.
There is also strong empirical support for this assumption in the national and regional
analysis performed in Latin American countries (Katz, 2000; Katz and Stumpo, 2001; De
Negri et al., 2005; Anlló and Suárez, 2008; Lugones et al., 2008; Suárez, 2009). The
incorporation of technology based on the acquisition of capital goods, the scarce
investments at the endogenous creation of knowledge, the low technological complexity of
the produced goods, and the bias towards exporting low added-value commodities explain
most of the innovative dynamic of Latin American firms. These findings, in turn, reinforce
the idea that if a firm persistently invests in innovation, then it is performing a different
innovative strategy—hypothetically, a more complex one. To put it bluntly, if a firm
invested in the incorporation of a new machine, it is more likely not to repeat the same
investment in the short term. If it does expend on innovation over the next period, this will
probably be further improving the productive process, incorporating new knowledge, or
performing complementary innovations. Once again, the continuity of the expenditure
seems to be positively associated with the complexity of the strategy, which means that the
more sustained over time the innovation expenditure, the more complex the innovative
strategy.
Chapter 4: Towards a conceptual framework for innovative strategies
- 144 -
Of course, a firm might continuously invest in the wrong innovation, might pursue an
innovation that ends up being useless because the environment has changed, might just
continuously invest low levels of resources in simple training or new basic equipment. In
those cases, the continuity of the expenditure will not be a good proxy for strategy
complexity—or, at least, it will be a fuzzy one. This is why, besides the continuity of
expenditure on innovation, the estimation of the complexity of the strategy has to be
complemented by an analysis of innovative structures and capabilities.
In this respect, the following methodological assumptions are made. The firm is pursuing
an innovative strategy if it has allocated resources to those activities (it has invested in
innovation). Firms that continuously invest in innovation have a continuous innovative
strategy, firms that stopped investing in innovation have a sporadic innovative strategy, and
firms that started investing in innovation have a new innovative strategy. If the firm did not
invest in innovation, then it is following a non-innovative strategy. Firms with a continuous
innovative strategy have a more complex innovative structure than the rest of the
strategies, where complexity is defined in terms of the relative level of investment in
innovation, the distribution of this investment among different innovative activities, the
relative level of capabilities, and the access to external resources to fund the innovation
projects. The higher these levels with regard to the sectoral average and after capital
ownership and size have been controlled, the more complex the innovative structure
(Chapter 5).
In this way, the first part of the motivation of this thesis (the existence of different firms)
can be translated into a methodological question (Q), the expected answers to which will
lead to the hypotheses, which will be discussed and empirically tested in Chapter 6.
Q1: Is there a positive relationship between the continuity of the innovative strategy and
the complexity of the innovative structure?
H1.1: Continuous innovative firms have a more complex innovative structure.
H1.2: New innovative firms changed their innovative strategy in the direction of a more
complex structure over time.
H1.3: Sporadic innovative firms have a low-complexity innovative structure sustained over
time.
Chapter 4: Towards a conceptual framework for innovative strategies
- 145 -
4.2.2. Innovative strategies and performance
The last part of the motivation of this thesis (firms better prepared to cope with the
environment) has to do with the impact of the innovative strategy on the firm’s
performance. In this respect, the main assumption is that continuous innovative firms will
have better performances. This has to do with the persistence of positive results
(innovations) and a greater impact of the results on the firm’s profitability, growth,
competitiveness, or a combination of these factors. Since continuous innovative firms are
supposed to have accumulated capabilities and resources, they are better prepared to deal
with the environment, meaning that they perform better.
The literature review shows that innovation results lead the firm to achieve extraordinary
incomes, which reinforce its competitive position and allow it to persist as a firm that
achieved innovation results (Nelson and Winter, 1982; Antonelli, 1997; Malerba et al.,
1997). Given the learning processes that every innovation triggers, firms that persist in
achieving innovation results will have higher dynamic capabilities for improving their
productive and organizational processes, identifying trends in demand, or moving quicker
than their competitors. As a result, these higher capabilities will increase the probabilities of
a better performance.
There have been three approaches to translating the concept of “innovations” into
empirical analyses: analyses based on technological innovations, those based on
technological and organizational innovations, and those based on patents. As innovation is
accepted as being an iterative, evolving process and the innovative process and its results
feed back into future innovative projects, the results will be interpreted in this study in their
broadest sense. This means that the innovation results considered will include product,
process, organization, or commercialization innovations at the firm level. Therefore, the
Oslo Manual definition of innovations (OECD, 2005) will be used here, based on the
arguments provided in 1.1.2.3.
In terms of impacts, the innovative strategy is expected to affect the firm’s economic
performance. The literature review shows that total factor productivity and market share
are the most accurate indicators for testing this relationship. Labor productivity is a second
best, but due to data restrictions, this is the option that will be used here, although
complemented by export and labor growth levels. The former is a proxy of the firm’s
Chapter 4: Towards a conceptual framework for innovative strategies
- 146 -
competitive position and its capabilities to escape from the domestic cycle, the latter a
measure of growth and changes in the productive scale.
In this case, I will assume that firms with a continuous innovative strategy are expected to
face additional learning processes and feedbacks that will allow them to increase their level
of resources and enhance their capabilities. This, in turn, will increase their probabilities of
persistently innovating and thus performing better. In practical terms, and following on
from the question regarding the innovative strategies, the last part of the main goal of this
thesis can be summarized as the following methodological questions, the corresponding
hypotheses of which will be discussed and tested in Chapter 7:
Q2: Is there is a positive relationship between the continuity of the firm’s strategy and its
past and present innovation results?
H2.1: The probability of innovating depends positively on having innovated in the past.
H2.2: The probability of innovating depends positively on the innovative strategy.
H2.3: The probability of innovating depends positively on having innovated in the
previous period when following a continuous innovative strategy.
Q3: Is there is a positive relationship between the continuity of the firm’s strategy, its
innovation results, and economic performance?
H3.1: Innovation results have a positive impact on firm performance (productivity, labor
growth and exports).
H3.2: Continuous innovative firms improve their performance improvements, regardless
of their innovation results.
4.3. Preliminary conclusions
The objective of this chapter was to propose a conceptual framework through which to
empirically approach the relationship between innovative strategy, innovation results, and
firm’s performance. To do so, empirical and conceptual literature was reviewed. The main
findings from the reviewed literature supports the idea that innovative structures at the
firm level can be analyzed based on innovation expenditure, human capital, and the firm’s
resources. However, the reviewed literature fell short on explaining how firms evolve in
terms of capabilities and how different innovative structures can be found beyond sectoral,
size, and even make versus buy characteristics. Even in those analyses where more than
Chapter 4: Towards a conceptual framework for innovative strategies
- 147 -
one period is considered, firms are analyzed in terms of their average behavior, thus
neglecting the possibility of changes in the innovative strategy and structure.
In this sense, the main conceptual contribution of this thesis is the presentation of a model
that accounts for the dynamic and evolving nature of the innovation process. The model
states that the innovation process can be linked to innovation results, and innovation
results to firm’s performance, but this double relationship has to be analyzed in dynamic
terms. By doing so, the inclusion of past actions and results allows for the consideration of
path dependence, and the possibility of a change in the strategy allows for the inclusion of
feedbacks and accumulation processes that arise not only from the innovative dynamic but
also from the firm’s productive performance.
The proposed conceptual approach constitutes an attempt to analyze firm behavior in
terms of the complex nature of innovation, so as to overcome the limits of static
“snapshot” approaches. This type of analysis can capture the innovative dynamic of firms
in a fixed time and place with a high level of precision, and can show how the situation
changed between two periods. However, they cannot explain the process that led the firms
from one situation to another. As I shall demonstrate, overlooking the dynamic aspect of
the innovation process distorts the empirical results and leads analysts to generalize
situations which are actually a matter of particular behaviors at the firm level and not
universal truths about the innovation process. The challenge for the next section lies in the
inclusion of innovation in dynamic terms, accepting the existence of micro-heterogeneity
and the possibility of a change in the firm’s innovative strategy.
Chapter 5: The dataset and methodology
- 148 -
Section III: Innovative strategies in the
Argentinean manufacturing sector
Introduction to Section III
The objective of this section is to apply the conceptual framework for innovative strategies
to the Argentinean case. Since these firms have been operating in an unstable environment,
the objective is to identify and characterize continuities and discontinuities in firm behavior
and, especially, changes in this behavior over time. To test the hypotheses, a model that
aims to measure the relationship between the innovative strategy, the innovation results,
and firm performance will be proposed.
The model will be applied to a panel of 800 Argentinean manufacturing firms that
participated in several rounds of the Argentinean Innovation Survey (ENIT) for the period
1998–2006. The model proposes a dynamic approach to the firm involving interaction
between the firm’s innovation process, results, and performance, accepting the possibility
of a change in firm behavior. The methodology will be based on a cluster analysis, where
changes among firm categories are allowed; and on regressive estimations, to measure the
strength and the sign of the relationship among the variables associated with firm behavior
and performance.
Results based on expenditure frequency show different innovative strategies, each with
innovative structures with different levels of complexity. While the environment was
changing drastically, some firms continued the same high-profile innovative behavior and
were able to successfully overcome the crisis and take advantage of the growth period.
These firms performed better than the average industrial levels and sectoral patterns.
Conversely, results also show some firms remained with the same non-innovative strategy
and also survived and grew, although their performances were worse. A third group of
firms radically changed their innovative strategy after the crisis and continued with high-
profile innovative behavior during the following periods. A fourth group shows a sporadic
innovative strategy, with isolated innovation investments and low capability levels.
Introduction to Section III
- 149 -
The first group of findings shows that some firms have successfully innovated in an
unstable environment. Sustained investments in innovation, qualified human resources,
combined efforts in knowledge, and intense interactions with the rest of the NIS explained
the high profile of continuous innovative firms. The second group illustrates that there is
no perfect macroeconomic environment or momentum of the economic cycle for fostering
innovation. The behavior of the third group of firms provides some insights regarding the
existence of path independence. These firms changed from a non-innovative to a highly
innovative strategy, a shift in trajectory that is best explained by high levels of capabilities
and interactions with the NIS. These two dimensions seem to be the source of firms’ path
creation.
In terms of impacts, firms with a continuous innovative strategy show higher probabilities
of achieving innovation results and higher performance levels, but a negative relationship
between past and present innovations (a negative persistence). A preliminary explanation of
this would be that these firms had to overcome the lock-in effect of innovations when the
environment changed, although their accumulated capabilities put them into a better
position to cope with the crisis and take advantage of the growth period. Those firms that
changed their strategy when the environment changed also present high probabilities of
innovation results and show a catching-up process in their economic performance after the
crisis. Finally, non-innovative firms were able to survive the macroeconomic crisis and
grow without investing in innovation for almost a decade, which is testimony to the
possibility of competing not being based on innovation. An intermediate performance is
observed among firms with sporadic investments, whose oscillations in performance level
seem to follow the economic cycle.
In order to present, apply, and discuss the results of the model, this section is structured as
follows. Chapter 5 includes a description of the dataset, the presentation of the model, and
the discussion of the methodology. In Chapter 6, the existence of different strategies is
tested and results are discussed. Chapter 7 consists of the analysis of the relationship
between the identified strategies and the firm’s innovative and economic performance. This
section also includes three statistical and methodological appendixes containing
clarifications about the dataset, additional tests, and complete estimations.
- 150 -
Chapter 5: The dataset and methodology
Introduction
The objective of this chapter is to present the dataset from which innovative strategies will
be identified and the treatment of the variables that will be used to construct the models
and to test the hypothesis. This is, of course, a descriptive chapter with no hypothesis or
research questions, but it is nonetheless very important in terms of data considerations and
the assumptions underlying it.
As I explained in the presentation of this section, the dataset consists of a panel made up
of 800 manufacturing firms that participated in several rounds of the ENIT. Although this
panel is not representative of the Argentinean industrial population as a whole, it
constitutes a unique source of information regarding the innovative behavior of different
size firms from different sectors, not only because of the number of cases it includes but
because of the possibility of merging variables so that the same firm is observed for more
than one reference period (over nine years, in this case).
This chapter has three main sections: the presentation of the data (5.1), the treatment of
the variables (5.2), and the integration of the conceptual framework and methodological
approach (5.3). At the end of this chapter, some preliminary conclusions are provided.
5.1. The dataset
5.1.1. ENIT - Argentinean Innovation Surveys
Argentina was one of the first Latin American countries to collect data about innovation
activities at the firm level.67 The use of sampling techniques based on the industrial census
allowed these exercises to be extrapolated to the manufacturing population, and the
application of the Oslo Manual led the ENITs to resemble the European Community
Innovation Survey (CIS) in their key modules and questions. As a consequence, results of
the exercises are directly comparable with international (especially European) results
67 The first ENIT was conducted in 1998 (INDEC, 1998) and covered the period 1992–1996. The second ENIT Second Innovation Survey was carried out in 2002 and gathered information for the period 1998-2001. (INDEC, 2003) The third survey was published in 2006 with information for the period 2002-2004. (INDEC, 2006) After 2006, the collection of statistical information on innovation became an annual exercise, but a decline in the agency’s budget worked against its wide distribution, and only the results for the year 2005 were published. (INDEC, 2008) Information for the years 2006 to 2010 has been collected, although it has not been published by the date this thesis was written. Appendix 1.1. summarizes the main methodological features of each survey.
Chapter 5: The dataset and methodology
- 151 -
regarding the main indicators of innovation in firms, despite being restricted to the
manufacturing sector.
In 2006, on the initiative of Argentina’s Ministry of Economy, the Enterprise Performance
Database (Base de Datos de Desempeño Empresarial, BDDE) was built (INDEC, 2007). This
database merged information from the ENIT with information from administrative
records and Industrial Surveys from 1998–2004.68 In 2010, the 2005 and 2006 ENIT were
processed and the BDDE was extended up to 2006 (INDEC, 2010). The database used in
this thesis originates in the BDDE and consists of 800 cases (approximately 50% of the
total number of firms included in each ENIT) with information on innovation inputs,
outputs, linkages, performance, and exports for 1998–2006 for different size manufacturing
firms from different sectors.69
These 800 cases are the result of merging the final rounds of the ENIT based on the firms
that participated in all four instances of the survey and responded to the questions about
sales, employment, and innovation activities. My selection of a smaller balanced panel is
partly due to the limits of the information that I was able to access (the BDDE is not a
public database). Furthermore, since I am interested in the dynamics of the innovative
strategy, a key element is observing firms continuously during an extended period of time
in order to analyze intra-firm changes in the variables. In other words, the same firms need
to be observed over different periods. Although statistical significance with respect to the
total population is lost, this does not undermine the panel’s ability to contribute to the
objective of this thesis, which is to analyze innovative behavior at the firm level over time
and during periods of macroeconomic instability. In the following section, this panel is
briefly presented and compared to the total population of Argentinean manufacturing
firms.
5.1.2. The panel and the population of manufacturing firms
5.1.2.1. Size: sales and employment
The panel is made up of 800 manufacturing firms that differ in terms of their size, capital
ownership type, and activity sector. Table 5.1 summarizes the distribution of the panel
68 At the time of the first ENIT (1998), the key innovation indicators had not yet been standardized and the exercise was carried out with an ad-hoc questionnaire, which means that much of the information is difficult to compare with the rest of the data. 69 The overall size of the BDDE for the period 1998–2006 is estimated at an unbalanced panel of 2400 firms with information from four innovation surveys (over 6500 observations).
Chapter 5: The dataset and methodology
- 152 -
according to firm size, measured by level of sales. To ensure the panel would be
comparable with the results of the ENITs, the size distribution was based on the average
level of total sales for the period 1998–2001, which is the criteria applied by the INDEC
for the second ENIT (1998–2001). As the exchange rate for the period 1998–2001 was
ARS 1=USD 1, the level of sales in ARS for the panel is the same in USD. The size
breakdown criteria for the ENIT comes from information published by the INDEC and
changes from one period to the next due to inflation adjustments and the use of variables
at current prices.
Table 5.1: Size distribution—Panel and ENIT (% of firms)
Panel ENIT 1998–2001 ENIT 2002–2004 ENIT 2005
Small Firms 80.4 75.9 77.6 72.1
Medium Firms 14.4 17.3 15.8 12.4
Large Firms 5.3 6.8 6.6 15.5
Total 100 100 100 100
Panel and ENIT 1998–2001 breakdown criteria: small firm sales < ARS 25 mill.; medium firm sales < ARS 100 mill.; large firms sales > ARS 100 mill.; ENIT 2002–2004: small firm sales < ARS50 mill.; medium firm sales < ARS 200 mill.; large firm sales > ARS 200 mill. ENIT 2005: small firm sales < ARS55 mill.; medium firm sales < ARS 220 mill.; large firm sales > ARS 220 mill. Source: Own elaboration based on INDEC (2003; 2006; 2008; 2010).
Around 80% of the panel is made up of small firms, which is 5 percentage points higher
than the average values observed in each ENIT. The difference in the participation of large
firms between the first and the last periods is explained mainly by the size breakdown
criterion. In 1998, a large firm had sales over ARS 100 million. After the devaluation of the
Argentinean peso in 2002, large firms became those with sales greater than ARS 200
million in 2002 and ARS 220 million in 2005, in all cases, measured at current prices.70 The
increase in prices was around 21% during 1998–2001, and 30% during 2002–200571,
therefore, the 10% adjustment on the criteria for 2005 (from 200 to 220) is insufficient to
compensate the effect of price increases. That is, “large firms” are smaller in 2005 than in
2002–2004.
5.1.2.2. Activity sector
Based on the international standardized industrial classification (ISIC), the sector
distribution of the panel is similar to the distribution of each ENIT (Table 5.2), with five
sectors accounting for 50% of the cases: food and beverage (18%), textile (10%), chemicals
70 Unfortunately, constructing a different breakdown for each ENIT is not possible, given the lack of available information. 71 Official exchange rate, based on INDEC figures and the World Bank’s estimate of inflation (WB, 2012).
Chapter 5: The dataset and methodology
- 153 -
goods (8%), metal products and machinery (14%), and the automotive industry (5%).
These sectors also explained 50% of sales for the whole period and around 60% of the
employment for the total sample, which is similar to the participation of these sectors
within the economy, where, in 2005, they accounted for 60% of the total industrial value
added and 61.5% of industrial employment (CEP, 2012).
Table 5.2: Sectoral distribution—Panel and ENIT (% of firms)
Panel
ENIT 98–01
ENIT 02–04
ENIT 05
15. Food and Beverage 17.8 20.9 21.2 21.9
16. Tobacco 0.4 0.5 0.4 0.5
17. Textiles 9.8 8.3 9.3 8.4
18. Wearing apparel; dressing and dyeing of fur 3.3 3.2 2.8 2.7
19. Leather; luggage, handbags, saddlery, harness, and footwear 3.0 2.7 2.7 2.4
20. Wood and of products of wood and cork (except furniture) 3.0 2.5 2.4 2.4
21. Paper and paper products 3.3 3.0 2.7 2.7
22. Publishing, printing, and reproduction of recorded media 5.1 5.0 5.3 5.3
23. Coke, refined petroleum products, and nuclear fuel 0.9 0.7 0.7 0.8
24. Chemicals 8.4 10.0 9.5 9.9
25. Rubber and plastics products 4.9 5.5 5.6 5.0
26. Other non-metallic mineral products 5.8 5.2 4.8 4.8
27. Basic metals 2.9 2.9 3.1 3.2
28. Fabricated metal products (except machinery and equipment) 5.1 5.0 5.2 5.3
29. Machinery and equipment n.e.c. 9.1 8.3 8.5 9.4
30. Office, accounting, and computing machinery 0.3 0.1 0.1 0.2
31. Electrical machinery and apparatus n.e.c. 4.1 3.8 3.9 3.5
32. Radio, television, and communication equipment 1.1 1.2 1.2 1.3
33. Medical, precision and optical instruments, watches and clocks
1.5 1.4 1.3 1.3
34. Motor vehicles, trailers, and semi-trailers 4.8 4.2 4.1 4.4
35. Other transport equipment 1.5 2.4 2.0 1.6
36. Furniture; manufacturing n.e.c. 4.3 3.1 3.2 3.1
Total 100 100 100 100
Breakdown criterion: ISIC. Source: Own elaboration based on INDEC (2003; 2006; 2008; 2010).
5.1.2.3. Capital ownership of the firm
The distribution of the panel according to the capital ownership of the firm also resembles
the distribution in the ENITs, although with a higher percentage of nationally owned firms,
which represent 85%, while 9% are owned by foreign capital, and the remaining 6%
consists of firms of mixed ownership (between 1% and 99% of foreign capital) (Table 5.3).
The mixed-ownership firms are, on average, 70% foreign capital and 30% national capital.
This polarization between national and foreign capital explains why the breakdown of
Chapter 5: The dataset and methodology
- 154 -
ownership is usually carried out solely in terms of national firms and firms involving
foreign capital, which is the criterion that will be applied in this thesis.
Table 5.3: Capital Ownership—Panel and ENIT (% of firms)
Panel
ENIT 1998–2001
ENIT 2002–2004
ENIT 2005
100% national capital 84.75 79.68 79.72 79.46
Between 1% and 99% foreign capital 8.88 11.02 9.40 12.00
100% foreign capital 6.38 9.30 10.88 8.54
Total 100 100 100 100
Source: Own elaboration based on INDEC (2003; 2006; 2008; 2010).
5.1.2.4. Innovation investments and results
The expected bias towards the most successful firms is observed in the innovative dynamic
of the panel as compared to the surveys. Table 5.4 summarizes the main indicators
regarding expenditure levels and activity types.
Table 5.4: Innovative dynamics—Panel and ENIT (selected indicators)
Panel ENIT
1998–2001 2002–2004 2005–2006 1998–2001 2002–2004 2005
Innovation expenditure 1.72 3.3 1.43 2.0 1 1.3
Expenditure on capital goods 0.96 1.46 0.8 1.3 0.6 0.83
Expenditure on R&D 0.29 0.75 0.25 0.22 0.2 0.2
Innovative firms 56 76.4 68.1 78.3 61.5 61.9
Firms that achieved innovations
58.8 47.125 53 60.6 50.8 41.7
Linkages 61.4 ND 45.25 74 ND 55 ND: no data available. Innovation expenditure (on capital goods/R&D): total expenditure on innovation activities (capital goods/R&D) to sales (%); Innovative firms: firms that reported having invested in
innovation activities regardless of the result (% to panel/ENIT); Firms that achieved innovations: firms that reported having achieved product, process, organizational, and/or commercial innovations (% to panel/ENIT); Linkages: interaction with any NIS agent while performing innovation activities, year 2005 corresponds to 2006 (% to panel/ENIT). Source: Own elaboration based on INDEC (2003; 2006; 2008; 2010).
The total expenditure on innovation is higher within the panel—except for the sub-period
1998–2001—with a significant difference for the period 2002–2004. Within this period, if
the top 10 firms in terms of absolute values were eliminated (50% of the absolute
expenditure on innovation within this sub-period), the relative investments in innovation
are still much higher within the panel (around 2.5% versus 1% in the survey). The rate of
innovative firms (firms with innovation expenditure, regardless of the results) is also higher
in all periods except 1998–2001, although the rate of firms that achieved innovations is
below the values for industry as a whole, except from 2002–2004.
Chapter 5: The dataset and methodology
- 155 -
5.1.3. The panel: cross-relationships and distribution
Given the impact of the size, sector, and capital ownership of the firms on the possible
strategies, firms’ capability to adjust their strategy, and the existence of technological and
competitive opportunities for innovation (Chapter 4.3), it is worth briefly characterizing the
panel in terms of these sources of micro-heterogeneity.
Firstly, given that the period to be analyzed starts on the first year of a macroeconomic
recession and ends on a year of significant economic growth, it is worth observing the
evolution of the panel in terms of firm size based on total employment, rather than sales,
to avoid the impact of inflation. In Table 5.5, the panel was classified into quartiles using
total employment in 1998, and average values for sales and employment were then
estimated. Given the evolution of prices during the period, as well as the 2002 devaluation,
average sales for every available year were deflated using the 1998 producer price index
(PPI) estimated by the INDEC.
According to the information provided in Table 5.5, smaller firms have average sales below
ARS 4.9 million (QI and QII), while larger firms’ sales are above ARS 17.6 million (QIII
and QIV). Since the largest firms (QIV) are around 10, 7, and 3 times larger than QI, QII,
and QIII, respectively, 50% of the panel can be said to be made up of small firms, 25% of
medium-sized firms, and the remaining 25% of large enterprises. The distance between the
two extreme quartiles and the rest of the sample is significant in terms of both sales and
employment, although it decreases by the end of the period. The distance in the
relationship between sales and employment is remarkably lower, which accounts for a more
homogenous panel in terms of productivity and, as will be discussed in Chapter 5.1.3, this
also accounts for the bias towards the most successful firms in terms of performance.
The evolution of sales and employment and the relationship between the two variables is
consistent with the macroeconomic evolution discussed in Chapter 2. All quartiles present
similar trends, with an increase in sales closer to 60% among the smallest firms (QI) and
around 40% among the larger ones (QIV) between the first and final years of the period. In
terms of employment, this increases among the smaller firms (QI and QII), remains
constant in medium-sized enterprises, and drops around 18% in larger firms. Finally, the
impact of the 1998–2001 recession is also evident, with a generalized drop in sales and
employment of approximately 20 percentage points for QII and QIII, whereas for large
firms (QIV), where sales remain constant, and small firms (Q1), where total labor increases.
Chapter 5: The dataset and methodology
- 156 -
Given the fact that values are measured in constant prices, both the growth and the
decrease are good proxies for the real evolution of the firms and provide a clear picture
regarding the size of the firms that make up the panel.
Table 5.5: Size evolution of the panel (breakdown by quartiles of employment)
1998 2001 2002 2003 2004 2005 2006
Sales (000 ARS)
QI 3595 3142 2803 3629 4410 5058 5657
QII 4880 3824 3591 4525 5397 5731 6893
QIII 17612 16011 15358 18548 21206 19954 24240
QIV 85420 82741 84579 93808 105438 110378 127012
Mean 27877 26430 26583 30127 34113 35280 40950
Employment (People)
QI 31 38 33 34 38 44 44
QII 54 49 46 48 53 67 64
QIII 120 106 91 97 115 137 126
QIV 567 504 309 317 476 551 474
Mean 193 174 120 124 170 200 177
Sales / Employment
(000 ARS)
QI 116 83 84 105 116 116 129
QII 90 79 78 95 103 85 109
QIII 146 151 169 191 185 146 192
QIV 151 164 274 296 222 200 268
Mean 144 152 222 243 200 177 232 QI, QII, QIII, and QIV: first, second, third, and fourth quartile, respectively. Source: Own elaboration based on INDEC (2010).
The cross-tabulation between the quartiles and activity sector of the panel (Table 5.6)
shows the expected bias towards the larger quartiles among scale-intensive sectors (such as
tobacco and paper) and a higher concentration of firms within the middle quartiles in
labor-intensive sectors (such as textiles). The food, chemical and metalworking industries
show similarly balanced distribution between the quartiles, which is consistent with the
heterogeneity of the Argentinean productive system discussed previously (Chapter 2.3) and
with the theoretical arguments regarding the weakness of the selection process to expel
some less productive firms from the market (Chapter 3.1.2).
Consistent with the correlation between foreign and large firms, the relationship with
industrial activity shows a higher participation of foreign firms in scale-intensive sectors,
while national firms are more concentrated in labor-intensive activities (Table 5.6).
Chapter 5: The dataset and methodology
- 157 -
Table 5.6: Sectoral distribution by quartiles and capital ownership (Panel=100 % of firms)
Panel by size Panel by capital ownership
QI QII QIII QIV 100%
national capital
1% to 99% foreign capital
100% foreign capital
15. Food and Beverage 21.1 19.0 26.1 33.8 88.73 7.04 4.23
16. Tobacco 0.0 0.0 33.3 66.7 33.33 33.33 33.33
17. Textiles 17.9 35.9 26.9 19.2 93.59 5.13 1.28
18. Wearing apparel; dressing and dyeing of fur 30.8 19.2 34.6 15.4 100.00 0.00 0.00
19. Leather; luggage, handbags, saddlery, harness, and footwear
20.8 25.0 29.2 25.0 95.83 0.00 4.17
20. Wood and of prod. of wood and cork (exc. furniture) 25.0 29.2 16.7 29.2 95.83 4.17 0.00
21. Paper and paper products 30.8 19.2 30.8 19.2 80.77 15.38 3.85
22. Publishing, printing, and reproduction of rec.media 31.7 24.4 22.0 22.0 85.37 9.76 4.88
23. Coke, refined petroleum products, and nuclear fuel 0.0 0.0 42.9 57.1 57.14 28.57 14.29
24. Chemicals 14.9 23.9 14.9 46.3 71.64 8.96 19.40
25. Rubber and plastics products 30.8 23.1 28.2 17.9 84.62 7.69 7.69
26. Other non-metallic mineral products 32.6 21.7 19.6 26.1 73.91 15.22 10.87
27. Basic metals 39.1 21.7 13.0 26.1 78.26 17.39 4.35
28. Fabricated metal products (exc. mach. and equip.) 24.4 31.7 24.4 19.5 85.37 9.76 4.88
29. Machinery and equipment n.e.c. 28.8 23.3 26.0 21.9 82.19 10.96 6.85
30. Office, accounting, and computing machinery 50.0 50.0 0.0 0.0 100.00 0.00 0.00
31. Electrical machinery and apparatus n.e.c. 12.1 33.3 42.4 12.1 84.85 9.09 6.06
32. Radio, television, and comm. equipment 33.3 33.3 0.0 33.3 55.56 33.33 11.11
33. Medical, prec. and optical inst., watches and clocks 50.0 16.7 25.0 8.3 100.00 0.00 0.00
34. Motor vehicles, trailers, and semi-trailers 21.1 26.3 26.3 26.3 71.05 15.79 13.16
35. Other transport equipment 33.3 41.7 16.7 8.3 100.00 0.00 0.00
36. Furniture; manufacturing n.p.c. 38.2 29.4 29.4 2.9 94.12 2.94 2.94
ISIC classification. QI, QII, QIII, and QIV: first, second, third, and fourth quartile, respectively. Source: Own elaboration based on INDEC (2010).
Finally, the cross-tabulation of size and capital ownership for the firms in the panel shows a
high correlation between foreign and large firms, while national firms are evenly distributed
across sizes (Table 5.7). This correlation will reappear later on (see 5.3), as the inclusion of
one of the factors will override the impact of the other in the econometric estimations.
Table 5.7: Capital ownership and size (Panel=100 % of firms)
QI QII QIII QIV Panel
100% national capital 27.3 27.7 25.2 19.8 100.0
Between 1% and 99% foreign capital 14.1 9.9 25.4 50.7 100.0
100% foreign capital 9.8 9.8 21.6 58.8 100.0
QI, QII, QIII, and QIV: first, second, third, and fourth quartile, respectively. Source: Own elaboration based on INDEC (2010).
5.1.4. Selection bias and representativeness
Several authors have explained the selection bias of innovation surveys caused by the fact
that firms that invest in innovation are more prone to answering such surveys, which lead
to overestimations of the role of innovation within the firm’s strategy as well as the impact
of the innovation investments on results—analogously, successful innovative firms are
Chapter 5: The dataset and methodology
- 158 -
more prone to responding than unsuccessful ones—(Raymond et al., 2007, 2010; Hervas
Oliver et al., 2012). Merged innovation surveys (a micro-data panel) have an additional bias
given by the higher survival rate among innovative firms.
The question of the representativeness of each survey aside72, the requirement of having
participated in all four of the exercises increases the selection bias of the sample, especially
given the macroeconomic situation of the reference period. This requirement also implies
that all firms considered were established before 1998 (first year of the first innovation
survey) and that they all survived the worst crisis in Argentina’s history. That means that
the sample is biased towards “successful” companies, or at least towards those who
managed to cope better with the recession of 1998–2001. The impact of this bias is a
possible overestimation of innovation results and economic performance, given the fact
that firm performance is assumed to be greater than the average level of the Argentinean
manufacturing industry.
All these biases, however, do not undermine the usefulness of the database, as it still allows
a relatively large group of firms (50% of the panel of each survey) to be studied over time,
within a period of deep macroeconomic changes. At the same time, since all the surveys
ask a large number of questions to both firms that invested in innovations and firms that
did not, behaviors can be analyzed by comparing several variables among different
innovative performances. Finally, the database also covers many sources of heterogeneity,
such as size, activity sector, and capital ownership, which allows different firm types to be
studied over a long period.
Table 5.8 summarizes the panel’s participation in terms of sales, exports, and employment
in comparison with the total manufacturing sector (expanded values from the results of
each ENIT). Firms from the panel account for around 23%, 19%, and 17% of sales for
1998, 2002, and 2005, respectively. Participation is lower in terms of employment, with the
panel accounting for 18%, 13%, and 15% for each period, respectively. The situation is
slightly higher for exports, with the panel explaining around 22%, 17%, and 15% for each
period. The comparison in terms of the relationship between sales and employment is
consistent with the expected bias: firms from the panel have higher levels of productivity,
especially during the crisis (1998).
72 The sample for each subject was based on stratification techniques, and forced inclusion and expansion factors were estimated based on the response rate so that the final panel for each ENIT can be extrapolated to resemble the industrial population (see Appendix 1.1).
Chapter 5: The dataset and methodology
- 159 -
Table 5.8: Representativeness of the panel (selected indicators)
1998 2002 2005
ENIT (a)
Panel (b)
(b)/(a) ENIT
(a) Panel
(b) (b)/(a)
ENIT (a)
Panel (b)
(b)/(a)
Sales 96.895 22.301 23.0% 169.947 33.272 19.6% 378.368 65.196 17.2%
Employment 821 152 18.5% 708 96 13.5% 1.034 160 15.4%
Sales/employment 118 144 1.22 240 347 1.44 366 408 1.11
Exports 17.147 3.764 22.0% 44.516 7.679 17.3% 114.776 17.059 14.9% ENIT: expanded values. Sales and exports in millions of current ARS, employment and sales to employment in thousands of people. Exports 2002 corresponds to 2003. Source: Own elaboration based on INDEC (2003; 2008; 2010)
The relatively high participation of the panel compared with the manufacturing sector as a
whole (800 firms accounting for 17% of total sales) can be explained by the characteristics
of these exercises within the Argentinean Statistical System. ENITs are included in the
national statistical system, so respondents are compelled to respond (Law 17622/1968),
which explains the 70% response rate for all the exercises. Although the mandatory
character of the survey improves the response rate, it can work against the accuracy of the
information provided by the firms. In the reverse situation, if the innovation surveys were
not mandatory, there would have been greater probabilities of their presenting a deeper
self-selection bias, as firms that achieved innovations are more predisposed to talk about
what they have achieved.
Another characteristic of the ENITs is that there has been no freely available information
about the disappearance of cases from one survey to the next. Since an original sample was
randomly extracted from the industrial population and then replacements for the drops
were also randomly selected from the population, the differences in cases between surveys
can be either the result of a non-response or the death of the firm. Therefore, the panel of
firms which responded to the four surveys does not account for the survival/death of
firms but for the behavior of a group of survivors.
The third characteristic of the Argentinean surveys is the different span of years included in
each exercise. Given the fact that some questions are asked regarding the period, the
information used in this thesis will be treated as a three-observation database, each one
associated with a specific sub-period, and average values will be estimated. As was
mentioned before, the 2006 ENIT has not been published yet and that is why it has not
been included in the previous sections. However, the results of this survey are available for
the panel of 800 firms and that is why it has been included in this explanation, combined
with the results for 2005, in order to extend the span of years and improve comparability
Chapter 5: The dataset and methodology
- 160 -
with the other two surveys. In other words, from now on the 2005 and 2006 surveys will
be treated as a single two-year survey.
5.2. The selected variables: definitions and treatment
5.2.1. Segmentation of periods and general data treatment
The panel covers a period of nine years (1998–2006) although these are unequally
distributed between the surveys into three sub-periods: 1998–2001, 2002–2004, and 2005–
2006. These coincide with the reference periods of the innovation surveys. As such, the
panel is made up of three observations for each case (firm). This segmentation also allows
distinctions to be made between different macroeconomic environments and for the
possible lags between innovation investments and results to be accounted for. This is due
to the fact that innovation is a process that may go beyond the calendar year and also to
data restrictions.
From these sub-periods, continuous variables were recalculated as annual averages, and
those in local currency were deflated by the producer price index, base 1998 at three digits
ISIC on a yearly basis. Given the exchange rate in 1998 (ARS 1=USD 1), the final variables
are expressed in constant USD. For dichotomous variables, the criterion was that of a
positive response in at least one year of the sub-period.
In the following sections, each variable is presented and linked to the conceptual work
presented in the previous chapters. In order to simplify the presentation of the variables,
their treatment, and the assumptions underlying them, descriptive statistics are presented in
Chapter 6 and 7, where the relationship between the different variables is also analyzed.
5.2.2. The innovative strategies
The four innovation surveys asked for the performance of 10 different innovation activities
(IA) in terms of the total amount expended in each year of the reference period. The
innovation activities are: internal R&D, external R&D, acquisition of capital goods,
hardware, software, technology transfer, training, engineering and industrial design, and
consulting.73 Respondents were asked about their performance of these activities during the
quest for innovations, regardless of the results.74
73 See Appendix 1.2 for definitions of each activity and the matching of variables. 74 The applied definition of expenditure explicitly excludes regular investments or any effort not related to the quest for new product, process, or organizational/commercial practices.
Chapter 5: The dataset and methodology
- 161 -
For the first step in the analysis of the strategies, variable innovation activities were
transformed into a dummy variable equal to 1 if the firm has expended on innovation
activities in at least one year within the sub-period, 0 if it has not (Iexp). The assumption
behind this variable is that if the firm invested in innovation, then it has decided to seek an
innovation-based competitive advantage, regardless of the type and level of investments.
The innovative strategies will then be classified according to the frequency of the
expenditure, based on the Iexp indicator over time. Between and , a firm can
choose to continue with its innovation investments, to start new investments, or to stop
them. Since the panel is made up of three periods, firm strategies will be classified
according to their behavior from t-2 to t-1 and from t-1 to t as follows:
Continuous innovative firms (Cont): firms with sustained investments in
innovation during two sub-periods;
Non-innovative firms (NI): firms with zero investments in innovation during two
sub-periods;
New innovative firms (New): firms which started to invest in innovation from one
sub-period to the following one;
Sporadically innovative firms (Spor): firms which stopped investing in innovation
from one sub-period to the following one.
Between t-2/t-1 and t-1/t, the firm can stay with the same strategy or change it, depending
on the evolution of the expenditures on innovation. In this way, a change in innovative
conduct is allowed for. Table 5.9 shows every possible combination and the resulting eight
situations75.
These strategies can be interpreted at two different levels. In theoretical terms, and
according to what was discussed in Chapter 4.2.1, continuous firms are those with a
strategy (the way they face competition) based on innovation where the feedbacks from
interaction with the environment and the implementation of the strategy reinforce the
behavior. The same behavior-reinforcing feedback process is assumed to exist among non-
innovative firms although with the opposite strategy, (not based on innovation).76 These
75 A detailed characterization of firms’ innovative strategy as well as their description in terms of the
innovation expenditures, capabilities, and resources (innovative structure) will be carried out in Chapter 6. 76 Given the period under analysis and the fact that capitalist competition forces firms to continuously seek a competitive advantage, an interesting line for future research is how these firms survive and what the elements of their strategy are that allow them to remain in markets where other firms compete based on an innovative strategy.
Chapter 5: The dataset and methodology
- 162 -
firms are assumed to have managed to survive the selection process and even grow because
this is affected by institutions that limit the extent of Schumpeterian creative destruction
and allow firms with lower productivity levels to remain in the market (Chapter 3.3.1).
Between these two strategies, new and sporadically innovative firms respond to the
following assumptions. Sporadic firms compete based on isolated investments in
innovation (problem-solving or short-term responses), but their main competitive strategy
is not based on innovation but other ways of survival which are to some extent closer to
those of non-innovative firms. New innovative firms, on the contrary, decide to modify
their behavior, and this change is associated with a new innovation-based competitive
strategy.
Table 5.9: Combinations of innovative strategies
Sub-period
Innovation expenditure
1998–2001 2002–2004 2005–2006
YES
YES Continuous
YES Continuous
NO Sporadic
NO Sporadic
YES New
NO Non-innovative
NO
YES New
YES Continuous
NO Sporadic
NO Non-innovative
YES New
NO Non-innovative
Yes (no) means that the firm declared (did not declare) innovation expenditure in at least one year of the sub-period. Source: own elaboration based on INDEC (2010).
From a methodological point of view, two extreme behaviors can be defined: firms that
invest in innovation and firms that do not. When these behaviors are analyzed in dynamic
terms, firms face two possible choices: to stop or to continue with the behavior.
Continuous and non-innovative firms both decided to continue with the same behavior,
sporadic and new innovative firms decided to change theirs. When these possibilities are
analyzed in terms of a specific period (in this case, 1998–2006), the combination of choices
gives the different possible outcomes; the eight innovative strategies.77 The literature on
77 The extension of the period under analysis exponentially multiplies the combination of choices and could enhance our comprehension of the evolving process of strategic decision, implementation, and adjustments and of course, our knowledge of firm behavior. This is another interesting line of future research.
Chapter 5: The dataset and methodology
- 163 -
innovative strategies only looks at the final decision—or the average choice—but it does
not analyze the transition from one situation to another (Chapter 4.1) and this is precisely
the intended conceptual and methodological contribution of this thesis.
The four strategies defined above are this author’s contribution to the literature, in the
sense that I want to analyze not only actual firm behavior but also the strategic decision-
making process and the extent to which the different combinations impact on innovative
structure and, ultimately, the firm’s performance.
From both the theoretical and the methodological points of view, we cannot know about
the choices made before the first sub-period (before 1998) and this is something that
cannot be solved by extending the database. To tackle this limitation, unobserved effects
will be included in the econometric estimations.
The second step in the characterization of the innovative strategies consists of the
characterization of the innovative structure based on the firm’s level and type of
investments in innovation, capabilities in terms of skilled human resources and linkages,
and access to external resources. The third step is related to the relationship between the
strategies and firm performance in terms of productivity, employment, and exports. In the
following sections, these variables are defined.
5.2.3. Innovation investments
Based on the type and level of investments in innovation, two new variables were
constructed: innovation intensity (II) and innovation balance (IB). To estimate innovation
intensity, the average innovation expenditure was divided by the average total sales for the
sub-period. The underlying assumption is that the higher the expenditure, the greater the
firm’s commitment to the quest for technological and organizational improvements; an
assumption widely confirmed in the reviewed literature (Chapter 4.1) and empirically tested
in other articles based on the same data but using different models and panels
(Chudnovsky et al., 2004; Lugones et al., 2007; López and Arza, 2008).
Innovation balance, or the combination of different innovation investments is another
indicator proven to have an impact on the innovative dynamic (Jensen et al., 2007; Lugones
et al., 2007; Yoguel et al., 2011). The argument indicates that the generation and application
of knowledge (or a new combination of existing knowledge) and its introduction into the
market in the form of innovations is the result of deliberate investments, which may be
Chapter 5: The dataset and methodology
- 164 -
more or less planned, in the pursuit of technological and organizational improvements (see
Chapter 3.3). These invesments must combine the incorporation of knowledge developed
outside the firm (exogenous knowledge) with the endogenous creation of new knowledge
in order to absorb and transform it into an innovation. Even in cases where innovation
takes place exclusively through the acquisition of embodied technology (capital goods), the
impact is greater when the firm also invests in activities to select, adapt, and improve it
(such as engineering and industrial design). To calculate this indicator, and adaptation of
the formula developed by Lugones, Suárez, and Le Clech (2007) was applied to the set of
innovation activities.78
The general notation of the index is:
∑|( ) |
Table 5.10: Innovation expenditure categories
Category (j) Description Weight (α)
A Research and development (internal and external) 0.25
B Engineering and industrial design + training 0.25
C Capital goods + hardware 0.25
D Technology transfer + consulting + software 0.25 Source: Lugones, Suárez, and Le Clech (2007).
Where i is the identifier of the firm, j is the identifier of each category of expenditure
summarized in Table 5.10, g is the expenditure in each category (j), AI is the cumulative
total expenditure on innovation activities, α is the weighting coefficient for each j (in this
case 0.25 for each set of activities), and n is the total number of categories analyzed (four in
this case). The cumulative expenditure will be used in each sub-period because it is
expected that the firm will distribute costs between the innovation activities based on
funding availability and the ability to spread the cost: for example, after the purchase of a
capital asset, investments in engineering and training to optimize its use should be allocated
and the latter will probably represent lower expenditure than the former but is not less
important in terms of the pursued innovation. Of course, this index is reduced to the group
of innovative firms (firms that reported spending on innovation, regardless of the results)
given its denominator. For non-innovative firms, the innovation balance was set to zero in
order to avoid the loss of observations.
78 Using a similar dataset for Argentinean manufacturing firms, the authors show that the group of “balanced” firms is the one with the best performance in terms of productivity and sales.
Chapter 5: The dataset and methodology
- 165 -
The index does not aim to find an optimal value for innovation balance, but to analyze
how different balances impact innovation results and firm performance and that is why the
weights of the categories are based on an “ideal type of balanced firm”. From a theoretical
point of view, the rationale behind assigning similar weights to each category (0.25) is that
it is not possible, a priori, to establish which type of activity is more important in terms of
the firm’s productive and competitive dynamic. In other words, it is to be expected that the
firm’s results and performance will be more impacted by a balanced strategy—one that
combines exogenous and endogenous investments—but I cannot establish assumptions
regarding the combination of specific activities (in some cases it could be capital goods and
training, in others engineering and hardware, or technology transfer and R&D).79
From Chapter 2 we know that the average behavior of the Argentinean firms is biased
towards the acquisition of capital goods, with low investments in endogenous forms of
knowledge creation (especially R&D). Therefore, firms with a higher innovation balance
combine different sources of knowledge in a more balanced way than the average level of
the panel. The closer to 1 the index is, the more balanced the expenditure is, the more
diversified the combination of sources of knowledge. Once again, the objective of this
index is to provide a single measure of distribution of expenditures. The IB index is a
diversity index that helps to summarize the combination of innovation investments (one
index instead of the several types of investments), and a positive relationship can be
assumed to exist between the IB and the complexity of the strategy in terms of the sources
of knowledge creation (Chapter 4.3). A firm with a high IB is expected to have a more
sophisticated innovative strategy than a biased one.
5.2.4. Innovation results
The four innovation surveys ask the firm if it has developed a new or significantly
improved product, process, organizational, or commercial technique during the reference
period. If so, the firm should state if this was new to the firm, the market, or the world. For
the purpose of this thesis, a firm will be said to have achieved innovation if it has declared
at least one innovation, regardless of the type (product, process, organization, or
commercialization) and the scope of the innovation (new to the firm, the market, or the
world). As a consequence, innovation results (Inno) will be measured using a dummy
79 Comparisons with different weights show marginal differences in terms of the index’s frequency. See
Appendix 1.3 for the complete estimations.
Chapter 5: The dataset and methodology
- 166 -
variable which takes the value 1 if the firm declared that it achieved at least one positive
result during the sub-period.
The theoretical rationale for this selection of innovation results in a broad sense (product,
process, organization, and commercialization) responds to the fact that this analysis is
concerned with the evolution of firm behavior from a dynamic perspective. In other words,
it is based on the recognition of the equal importance of technological and non-
technological innovations when it comes about the firm’s performance (Le Bas et al., 2011;
Lugones and Suarez, 2010; Nelson and Winter, 1982) and the fact that one specific
innovation could trigger new ones in other areas of the firm (a new product impact creating
a new commercial channel). It also allows us to acknowledge that if a firm has developed a
new product, it will probably invest in the exploitation of that product during the following
periods, which could demand process, organizational, or commercial innovations.
Since the relationship between results and investments will be tested, another assumption
has to be made. One widespread consistency test narrows innovation down to only those
firms that invest in innovations (innovative firms). However, since the time that elapses
between investments and results cannot be anticipated, for the purpose of this thesis, a
firm could achieve innovation results, without being innovative (positive expenditures).
The assumption behind this statement is that either the expenses were incurred during a
period not included in the surveys, or they were made with expenditures associated with
the ordinary operations of the firm.80
Finally, although the different number of years included in each sub-period could affect the
reported number of innovations (it is more likely that a company has innovated when the
consultation is for a period of four years than when it is asked for a period of two), the way
this variable is addressed in the questionnaire (in the four consultation exercises, “achieved
innovations” relates to the period and not to the year) makes it impossible to use variables
that span equal number of years.
5.2.5. Qualified human resources
Each innovation survey asks for the firm’s total number of employees and their
distribution by level of formal education. The options provided by the questionnaire are:
80 Even if the criterion of positive expenditures were applied, the results would not change greatly since there is a high correlation between innovative firms and firms that achieved innovations: 98% of the firms that had at least one positive result during 1998–2006 did also invest in innovation in at least one of the years.
Chapter 5: The dataset and methodology
- 167 -
basic education or less (secondary school), technical education (tertiary education), and
professionals (five years or more of university education).
The endowment of skilled human resources is a widely used measure of firm’s capabilities
(Chapter 4.1), usually estimated by means of the relative number of skilled employees
(Crépon et al., 1998; Raymond et al., 2010; Yoguel et al., 2011; Srholec and Verspagen,
2012). The assumption is that the greater the number of years of formal education, the
higher the average skills levels within the firm, and therefore the higher its capabilities level.
This variable was estimated as the annual average number of skilled personnel (university
degree) for each sub-period and then transformed into relative terms by dividing it by total
employment. As such, QHR is a continuous variable that accounts for the participation of
highly qualified labor.
5.2.6. Linkages with the National Innovation System
The firm’s linkages with the NIS are another variable usually included as a proxy for
capabilities, assuming that interactions with other organizations increase the firm’s
competencies (Chapter 1.2.2.4 and Chapter 4.2). The ENIT gathers information regarding
the interaction of the firm with other NIS agents during the process of seeking
innovations. Unfortunately, the information collected for the period 2002–2004 is not
comparable with the other two exercises (a different question was used), so it had to be
excluded from the analysis.
For the two comparable surveys (1998–2001 and 2005–2006), respondents were asked to
inform of interactions with universities, public organizations (includes funding offices and
programs), related firms (headquarters and subsidiaries), other firms, suppliers, clients,
R&D laboratories, and technological centers (public and private). Both surveys also asked
for a set of objectives: information, training, technical assistance, design, and R&D.
Using this information, two sets of variables were created. The first was a dummy variable
equal to 1 if the firm reported interactions with other NIS agents and zero if it has not
(Link). Given the gaps in the information, this variable was included as two time-invariant
dummies, one for the period 1998–2001 and one for the period 2005–2006, and they will
be activated for the corresponding period (multiplied by T).
The assumption behind this variable is that to generate new knowledge and transform it
into an innovation, the firm must access external knowledge. Despite the disappointing
Chapter 5: The dataset and methodology
- 168 -
empirical results (low cooperation rates), evidence shows that interactions happen when the
firm has crossed a threshold of minimum competencies (Tether and Swann, 2003; Erbes et
al., 2004; Suárez, 2009; Yoguel et al., 2011). For instance, a company that engages in R&D
is more likely to interact with R&D laboratories than a company that does not. For the
purpose of the present analyses, if the firm has established linkages, then it has crossed that
minimum threshold of competencies and has higher skills than a firm which has not
interacted.
The second variable is a set of dummies which will account for the firm’s different linkages
with the rest of the NIS in terms of organizations and objectives. This variable will be used
only for descriptive purposes since all interactions are assumed to have the potential of
generating complex knowledge, which is consistent with the literature on the NIS
(Lundvall, 1992; Narula, 2003) and the different ways of learning (Jensen et al., 2007;
Suárez, 2009).
5.2.7. Financial resources
The four innovation surveys ask about access to external resources for funding innovation
activities. Respondents are asked to distribute the declared expenditure between a set of
funding organizations: banks, suppliers, clients, other firms, public organizations, non-
governmental organizations, international organizations, universities, and other sources.
Funding via related firms was excluded.
Using this information, a binary variable was created for each sub-period that takes the
value 1 if the firm accessed external resources and 0 if it has not (ER). Since the lack of
resources has been signaled in all surveys as the most important obstacle to the realization
of innovations, the assumption behind this variable is that if the firm has managed to
overcome that barrier it will be in a better financial position. This better financial condition
will be interpreted as a better situation in terms of resources generally speaking. Of course,
one could argue that if the firm is in a good financial position, then it is unlikely to access
external funds (especially given the high interest rate), and if it does, in this case, better
financial situations will somehow be punished. The evidence, however, does not support
this hypothesis (Chudnovsky et al., 2004; Chudnovsky et al., 2006; Porta, 2006).
Chapter 5: The dataset and methodology
- 169 -
5.2.8. Performance: productivity, employment, and exports
Three performance indicators were constructed: productivity, growth, and exports. The
advantage of testing performance in different ways is that it will allow innovation to impact
differently on different performance dimensions. The theoretical support for this
approximation is based on the idea that not all innovations aim to create the same
competitive advantages: some firms could be seeking to increase their market share
(growth), others to increase their profit levels or minimize costs (productivity), while others
could be trying to access external markets (exports). Consequently, by analyzing these
different dimensions separately, we will be able to observe these different scenarios and the
different interactions between the innovative strategy and the firm’s performance.
Productivity (Pd) was calculated as the coefficient between total sales and employment
(labor productivity), in natural logarithm. The surveys do not ask about capital stock or
investments, so it is not possible to estimate more robust indicators of productivity (such
as total factor productivity). In any case, there is wide theoretical and empirical support for
this variable as a proxy for productivity (Kemp et al., 2003; Raymond et al., 2010; Yoguel et
al., 2011). As with innovation expenditure, sales were deflated using the PPI base 1998 and
calculated as average values for the sub-period.
The level of employment will be used as a proxy for growth (Gr), also calculated as annual
average values for the sub-period in natural logarithm. Given the change in the relative
prices (the 1998–2001 devaluation), labor seems to be a more accurate indicator than sales
to account for the firm’s growth in terms of size.
This analysis of labor growth in relation to the innovative strategy will also allow us to
analyze the hypothesis about technological unemployment (Coad and Rao, 2011). At the
firm level, the direct impact of innovations could be an increase in employment levels due
to the firm’s better competitive position and, therefore, its sales levels (an increase in
production). However, it could also cause lower employment levels, depending on the
productivity increases associated with the innovation (the replacement of labor by
machinery).81 In this case, the assumption is that innovations allow firms to expand their
activities and, therefore, to increase the employment levels. Of course, since this study will
also look at the evolution of productivity, and draw some preliminary conclusions about
81 The former case is usually associated with product innovations and the latter with process innovations. Analyzing the differential impacts of different types of innovations is beyond the objective of this thesis but it seems an interesting line of future research, especially in the case of Argentina, where besides the sustained growth of the last decade, there is evidence of structural unemployment (see Chapter 2.2.3.6).
Chapter 5: The dataset and methodology
- 170 -
the relationship between the innovative dynamic of the panel and technological
unemployment (the implementation of labor-saving innovations).
The impact on the firm’s competitiveness will be analyzed in terms of export levels,
calculated as the annual average value for the sub-period in natural logarithm (Xpo). The
use of absolute values (instead of relative measures such as the ratio to sales) will allow us
to avoid the impact of the increase in the domestic prices (because of changes in the
exchange rate and inflation), which distorts the real evolution of foreign sales. Of course,
environmental variables will allow us to control for size and sectoral belonging, two of the
most important factors affecting the absolute levels of exports. An additional control will
be included, in order to differentiate exporting from non-exporting firms (Xr). As such, Xr
is a binary variable equal to 1 if the firm has exported in at least one year of the sub-period,
and 0 if it has not.
From a theoretical perspective, exports can be included as a cause or consequence of the
innovative process (Molina-Domene and Pietrobelli, 2012). In the first case, the
assumption is that the exposure to competition in foreign markets requires the firm to
invest in innovation in order to keep up with the rhythm of the international technological
frontier, to compete in more sophisticated markets, or to differentiate its products in such
a way that consumers are willing to pay a higher price for them. Exporters thus face
differential incentives (compared with non-exporters) that affect innovation decisions
(Bernard et al., 2007). When exports are understood as results, the underlying assumption is
that the innovation process could lead to lower productive costs or differentiated products
which allow the firm to access to new markets or niches, including foreign ones (Porta,
2006; De Negri et al., 2005).
For the purposes of this thesis, and given the low foreign presence of Argentinean
manufacturing sector (Chapter 2.2), the relationship between innovations and export
performance will be understood as the impact of innovative strategies on exports.
5.2.9. Environmental characteristics: size, sector, capital ownership, and sub-
periods
To control for structural characteristics, size, sector, capital ownership and time controls
were included. The size control refers to the firm’s total employment measured as the
average annual value for each sub-period, in total number of people (Size). Since growth
Chapter 5: The dataset and methodology
- 171 -
(Gr) will be approximated using the same information, the size control when analyzing
performance it will include total sales, as average values for the sub-period.
Sectoral controls are based on the ISIC Revision 3.1, at two-digit desegregation. For the
regressions, one dummy will be included for each sector (Sector). For the descriptive
statistics and the cluster analysis, values will be tested in terms of the coefficient between
the firm’s level and the average level of the sector it belongs to. In these cases, results have
to be interpreted as relative levels compared with the sectoral average (the average level for
the whole panel is 1). The type of sectoral control will be specified in each case.
A “company with participation of foreign capital” is defined as a firm with more than 1%
of its shares owned by foreign capital in 2002–2004 (KO). As was presented before, the
average level of participation is high enough to control the specificities that arise from this
feature of the firm with only one dummy (70% of the capital of mixed ownership firms
belongs to foreign agents). At the same time, there are no significant differences between
firms with 100% capital ownership and those where it is between 1% and 99% for the case
of sales, exports, employment, and innovation investments (see appendix 1.4).
The last set of dummy variables refers to the impact of macroeconomic situation. Given
the available information, the panel was split up into three sub-periods: 1998–2001
(hereafter, t-2), 2002–2004 (t-1), and 2005–2006 (t). A dummy variable for each sub-period
was created (t).
5.3. Integration of the conceptual and methodological approaches
Table 5.11 summarizes the variables that will be used to test relationship between the
strategies, the innovative structure, the innovation results, and the firm’s performance
according to the conceptual framework (Chapter 4).
With i representing the firm and t the sub-period, the innovative structure for each strategy
will be analyzed in terms of innovation intensity (II), innovation balance (IB), the level of
qualified human resources (QHR), interactions with the NIS (Link)82, and access to external
funds to finance innovation activities (RE). To analyze the success of the innovative
strategies, variable innovations (Inno) will be used. The basic idea is that successful
82 Although QHR and Link are strictly related to the firm’s innovative behavior, both variables can be positive even if the firm did not invest in innovation (they can do innovative activities without monetary investments) so they will allow to control for different capability levels even if among non-innovative firms. The hypothesis, of course, is that both variables will be higher (or they will increase more) among innovative firms.
Chapter 5: The dataset and methodology
- 172 -
strategies allow the firm to innovate persistently and this allows the firm to sustain or
improve its market share or quasi-rents. Productivity (Pd), growth (Gr), and exports (Xpo)
will be tested against both the innovative strategy and the innovation results. In this way,
not only will the impact of innovations on performance be analyzed, but also the impact of
the learning process that arises from innovation activities, regardless.
Table 5.11: Summary of the dimensions and variables
Label Detail Calculation Value
Innovative Strategy
Innovation expenditure >0 at least one year of the sub-period 0;1
Continuous innovative firm >0 in t & t-1 / t-1 & t-2 0;1
New innovative firm >0 in t / t-1 0;1
Sporadically innovative firm >0 in t, t-1 / t-2 0;1
Non-innovative firm =0 in t & t-1 / t-1 & t-2 0;1
Innovative Structure
Expenditure on innovation
Innovation intensity Innovation expenditure to sales. Annual average of
the sub-period. 0–
Innovation balance Distribution of the total innovation expenditure. 0-
Capabilities
Qualified human resources Number of skilled human resources to employment.
Annual average for the sub-period. 0–
Linkages At least one linkage in the sub-period. Variable
available for t and t-2. 0;1
Resources
Access to external sources of funding At least one source of funding in the sub-period. 0;1
Innovation Results
Product, process, organization, or
commercialization innovations At least one innovation in the sub-period. 0;1
Performance
Productivity Total sales to employment, annual average for the
sub-period. In natural log. 0–
Growth Total employment of the firm, annual average for
the sub-period. In natural log. 0–
Exporter firm Exports > 0 in at least one year of the sub-period. 1;0
Exports Total exports, annual average for the sub-period.
In natural log. 0–
Environmental determinants (structural variables)
Total employment/sales Total employment/sales of the firm, average value
of the sub-period. 0–
Capital ownership A firm that has more than 1% of shares owned by
foreign capital in t-1 0;1
Sectoral belonging, 2-digit ISIC in t-1. One dummy for each sector. 0;1
Sub-period One dummy for each sub-
period. 0;1
Sub-indexes
Case Manufacturing firm with more than 10 employees. 1–800
Sub-period = 2005–2006; = 2002–2004; =
1998–2001 3
Chapter 5: The dataset and methodology
- 173 -
Table 5.12 shows the correlation matrix between the variables that will be used to test the
strategies. Although most of the variables for the innovative structure show an association
below 0.3, the positive and significant correlations account for the importance of analyzing
them together and for testing the existence of different combinations (an underlying vector
of means) for different groups of firms. The positive association between innovative
expenditure and the rest of the variables reinforces the conceptual idea of innovative
strategies associated with the frequency of expenditures and the type of investments and
human resources allocated to the quest for innovations (Chapter 4.2). Another factor that
confirms the earlier statement about the difficulties of financing innovation in the
Argentinean case is the low association between the level of expenditure (II) and access to
external resources (ER), even during the growth period (t). Finally, there is the relationship
between innovation balance (IB) and linkages with the national innovation system (Link).
This correlation can be explained by the fact that firms with a diversified investment
structure—a high IB—are firms that perform different innovation activities (R&D, capital
goods, engineering, training, etc.) so they need to combine different sources of knowledge
in order to both carry out the activities and internalize and apply the acquired knowledge.
Table 5.12: Innovative structure—Correlation matrix
t-2 t-1 t t-2 t-1 T t-2 t-1 T t-2 t t-2 t-1 T
t-2 1
t-1 0.18 1
T 0.11 0.29 1
t-2 0.21 0.12 0.12 1
t-1 0.10 0.30 0.21 0.42 1
T 0.10 0.19 0.22 0.38 0.61 1
t-2 0.06 0.22 0.13 0.27 0.20 0.20 1
t-1 0.00 0.34 0.11 0.04 0.14 0.09 0.46 1
T 0.03 0.19 0.10 0.18 0.21 0.17 0.61 0.44 1
t-2 0.16 0.13 0.12 0.44 0.29 0.25 0.23 0.09 0.18 1
T 0.10 0.21 0.24 0.23 0.44 0.47 0.20 0.10 0.20 0.22 1
t-2 0.29 0.06 0.08 0.33 0.16 0.12 0.05 0.00 0.02 0.26 0.05 1
t-1 0.01 0.17 0.15 0.06 0.24 0.15 0.05 0.13 0.07 0.05 0.14 0.11 1
T 0.06 0.09 0.31 0.16 0.20 0.23 0.08 0.10 0.06 0.13 0.18 0.12 0.34 1 Light gray: non-significant correlations; black: significant at 99%. Pearson’s correlations. Obs.: 800. Source: own elaboration based on INDEC (2010).
The methodology will therefore be based on descriptive and econometric techniques that
aim to measure these preliminary correlations and also the levels and signs of the
relationships between innovative strategy, innovative structure, the innovation results, and
the firm’s performance. In order to study innovative strategies and structures, descriptive
and statistical analyses will be performed first, and a cluster analysis will be run afterwards
Chapter 5: The dataset and methodology
- 174 -
(Chapter 6). Despite the limitations of the descriptive statistics used and the lack of
controls, they are still a good way of acknowledging relationships between variables and
identifying preliminary correlations. In this sense, median tests will be used to check the
strengths of relationships and establish main associations.
After the descriptive analysis, a cluster analysis will be run. One of the most quoted
definitions of the cluster analysis is that provided by Kaufman and Rousseeuw (1990), who
define this technique as the “art of finding groups in data,” meaning that the interpretation
of the results is strongly dependent on the analyst.
The standard procedure for cluster analysis starts with a factor analysis of several variables
in order to find significant latent variables (factor scores) which account for similar paths in
the levels of variables among cases (Blackman et al., 1973). Using these factor scores, the
cluster procedure groups cases based on measures of distance (similarity measures), so
between distances are maximized while within distances are minimized. This is usually an
exploratory analysis in which the hypotheses to be tested are usually based on the variables
used and not on predicted groupings.
In this thesis, variables and groups are part of the hypothesis, so a confirmatory analysis is
required, and a discriminant cluster technique was chosen. Similar to factor analysis, this
technique finds latent variables within a matrix and tests the similarity between the
provided grouping (in this case, the strategies) and the estimated one, also based on tests
over the between and within variations (Fix and Hodges, 1951). One of the advantages of
this technique is that it does not force the case within a group but estimates a case’s
probability of being correctly assigned to a group with similar means for a set of variables.
That is, a common vector of variables within each group. Another distinctive feature of the
model is that it allows the inclusion of continuous and binary data (Klecka, 1980; Tu and
Han, 1982), so all the variables selected as proxies for the innovative structure can be
jointly analyzed.
To combine the three dimensions of the analysis (cases, variables, and time), the panel data
was transformed into a cross-sectional dataset where each variable accounts for an
innovative characteristic at a specific time, for the 800 cases of the panel. As a
consequence, even though the transformation of the dataset eliminates time as a variable,
the results have to be read by looking at the evolution of sets of three interrelated variables
(one for each sub-period).
Chapter 5: The dataset and methodology
- 175 -
The next step is to include serial correlation and analyze the relationship between the firm’s
strategy and its performance (Chapter 7). The correlation matrix for the results and
performance variables is presented in Table 5.13. In this case, the serial correlation of
variables is high, which is consistent with the path dependence discussed in Chapters 3 and
4. Looking at the between correlations (among performance variables), the strongest
association is that observed for the case of exports and productivity in t-2 and t-1, an
aspect that is consistent with the literature discussed in Chapter 4.1. The association is
lower for the period t, which is also consistent with the change in the general
competitiveness of the industry given by the new relative prices (Chapter 2.3). A plausible
hypothesis here is that during the crisis and the recovery period, exports depended on the
firm’s performance. After the devaluation, firms were able to export not only because of
their productivity levels but because of the competitive shock of the new relative prices
(the new exchange rate). Finally, the correlation between innovations and the different
measures of performance is lower than expected and below 0.30 for all variables in all
periods. This lack of strong association contradicts what was expected by the literature and
the findings for other countries (Chapter 4.1) and will be analyzed in greater depth in
Chapter 7, where this correlation will be tested in a more robust and complex way.
Table 5.13: Innovation results and performance—Correlation matrix
t-2 t-1 t t-2 t-1 t t-2 t-1 T t-2 t-1 t
t-2 1
t-1 0.18 1
T 0.32 0.20 1
t-2 0.10 0.09 0.07 1
t-1 0.12 0.09 0.09 0.90 1
T 0.12 0.08 0.08 0.78 0.91 1
t-2 0.19 0.12 0.19 0.07 0.12 0.13 1
t-1 0.21 0.13 0.21 0.10 0.12 0.13 0.96 1
T 0.23 0.15 0.24 0.14 0.15 0.13 0.92 0.96 1
t-2 0.08 0.09 0.08 0.58 0.66 0.76 0.29 0.30 0.31 1
t-1 0.08 0.08 0.08 0.60 0.68 0.78 0.25 0.26 0.27 0.99 1
T 0.10 0.10 0.10 0.68 0.69 0.72 0.29 0.33 0.40 0.81 0.79 1 Light gray: non-significant correlations; black: significant at 99%. Pearson’s correlations. Obs.: 800. Source: own elaboration based on INDEC (2010).
The impact of strategies on firm performance will be analyzed by means of dynamic OLS
regressions with the data set back into a panel database, where variables, groups, and time
will be jointly analyzed. Unobservable effects will be also included. As was presented in the
conceptual framework (Chapter 4.2), the innovative strategy is just one of the dimensions
of the productive dynamic of the firm, so the model will partially explain firm performance
Chapter 5: The dataset and methodology
- 176 -
(for example, the productive and commercial processes, not included here, also determine
firm performance). At the same time, what is being studied is the firm’s behavior during a
narrow period, which implies that many of its past actions are omitted. Finally, there is a set
of specificities given by the managerial characteristics of the firm, the particular
combination of human resources and other characteristics, usually referred as the
idiosyncrasy of the firm. Therefore, the analysis of the interaction between the strategy, the
structure, and the performance has to include unobserved effects.
To analyze the results of the innovation process (the innovations), and since the dependent
variable is a binary one, a probabilistic model will be run. Performance indicators
(productivity, labor growth, and exports) will be tested by means of a dynamic regression.
The first estimation will allow us to know the probability that a firm will persist with
innovation. The second estimations will shed light on the association between the
innovative strategies and the firm’s performance.
Besides the sign and relative level of the coefficients estimated with the regressions, the
analyses of the firm’s persistence and performance will also include the interpretation of
the marginal effects of the independent variables (strategy and structure) on the dependent
ones (innovations, productivity, labor growth, and exports). Average marginal effects
account for the average relationship between the dependent and independent variables for
the average firm and if the rate of increase is constant. If the independent variable is a
dummy, then average marginal effects have to be estimated in terms of a positive response
(the dummy at value 1), and the effect refers to the differential impact of this variable in
those cases where the condition is satisfied (dummy=1) (Bartus, 2005). In this respect,
marginal effects must be analyzed with caution to the extent that they probably refer to a
non-existent firm on a theoretical path of growth of the independent variable. With regard
to this, although the interpretation of the level and sign of the coefficients is usually simple
and direct, the analysis of the marginal effects provides only an approximate preliminary
interpretation of the findings. However, analyzing the marginal effect allows a better
understanding of the coefficients and the intensity of the relationships between variables,
enabling us to extract summarized conclusions about the relationships.
5.4. Preliminary conclusions
The objectives of this chapter were to present the database to be used for testing the
hypotheses and to transform the conceptual framework into a methodological design
Chapter 5: The dataset and methodology
- 177 -
capable of corroborating—or not—the existence of different strategies with different
impacts on the firm performance. To do so, in the first section the database was described.
The innovative strategies will be tested using a balanced panel made up of 800 firms that
participated in four consecutive national innovation surveys, with information about
economic performance, innovative dynamics, and interactions with other NIS agents.
Despite the selection bias of the panel, it constitutes a unique source of information about
firms’ innovative behavior and performance for an extended period of time, including
different points in the economic cycle.
Following the description of the panel, the treatment applied to the selected variables was
presented in order to discuss how these variables should be interpreted. Three types of
variable will be considered in the model: those relating to the firm’s innovative dynamic
(strategy and structure), those that account for economic performance (productivity,
growth, and competitiveness), and those about results (innovations)..
After the treatment applied to each variable had been described, the conceptual framework
was integrated with the selected variables and the empirical approximation to the
hypothesis. As a result of this methodological exercise the strategies were stylized and the
concept of the innovative structure was transformed into observable features of the firm:
innovation expenditures were measured in terms of levels and balance, capabilities in terms
of the participation of qualified human resources and linkages, and the availability of
resources in terms of access to external funds. The firm’s performance will then be
approximated using innovation results, labor productivity, employment growth, and
exports.
Finally, a preliminary discussion regarding the statistical and econometric techniques that
will be applied was provided. The model aims to test the association between the strategies,
the innovative dynamic, and the firm’s performance. In this respect, cluster analyses and
dynamic regressions will be used. Cluster analyses will allow the strategy to be tested using
a confirmatory approach, so the level of accuracy of the classification will be provided
(Chapter 6). The regression will allow the simultaneous inclusion of cases, variables, and
time to test the impact of the innovation process on results and performance (Chapter 7).
- 178 -
Chapter 6: Regularities and changes in firms’
strategies
Introduction
The objective of this chapter is to test the existence of different strategies. The main
research question is how the concept of innovative strategies can improve our
comprehension of micro-heterogeneity. The hypothesis states that there is a positive
relationship between the continuity of the strategy and the level of complexity of the
innovative structure (relatively higher innovation investments, capabilities, and access to
external resources). Then, the taxonomy of strategies presented in Chapter 5 will be tested
on 800 innovative and non-innovative Argentinean manufacturing firms for an unstable
period (1998–2006) based on a cluster analysis.
The findings corroborate the hypotheses. The results show that the classification of firms
according to the frequency of their innovation expenditure leads to groupings of firms with
similar innovative strategies: continuous firms present high levels of complementary
investments in innovation and capabilities, which are sustained over time; new innovative
firms seem to have caught up in terms of investments based on pre-existent high levels of
capabilities; and sporadic firms show opportunistic behavior, based on short-term process
innovations, which are associated with the incorporation of machinery. Non-innovative
firms show the poorest results and a consistent non-innovative behavior over time.
This chapter is structured as follows. After this introduction, in section 6.1 the hypotheses
are defined and the methodology is presented. In section 6.2, the classification of the
strategies will be applied to the Argentinean panel, and the innovative strategies will be
identified and characterized. The cluster analysis is estimated in section 6.3, and the
findings from this are discussed in 6.4. Finally, some preliminary conclusions are provided.
6.1. Hypotheses and Methodology
6.1.1. Hypotheses
According to the strategy based-approach (Chapter 3), firms face the selection process by
following different strategies, and when the quest for a competitive advantage is based on
innovation, then firms are pursuing an innovative strategy. Firms can also seek a
Chapter 6: Regularities and changes in firms’ strategies
- 179 -
competitive advantage through a non-innovative strategy. The multiplicity of possible
strategies contributes to explaining the existence of micro-heterogeneity and understanding
how firms deal with the environment. The first step is then to study the existence of
different strategies in terms of more or less dynamic innovative behavior.
In Chapter 4, I claimed that innovative strategies can be analyzed by looking at the firm’s
frequency of innovative expenditure which is associated with different innovative
structures, which led to the formulation of the methodological questions (Q). In Chapter 5,
I identify four innovative strategies (continuous, new, sporadic, and non-innovative firms),
based on the frequency of expenditure on innovation. From a dynamic perspective (t and t-
1), this leads to eight possible combinations, which are set out in Table 6.1.
Table 6.1: Combinations of innovative strategies
Sub-period Reference
Innovation expenditure
1998–2001 2002–2004 2005–2006
YES
YES Continuous
YES Continuous
C-C
NO Sporadic
C-S
NO Sporadic
YES New
S-N
NO Non-innovative
S-NI
NO
YES New
YES Continuous
N-C
NO Sporadic
N-SP
NO Non-innovative
YES New
NI-N
NO Non-innovative
NI-NI
Yes (no) means that the firm declared (did not declare) innovation expenditure in at least one year of the sub-period. Source: own elaboration based on INDEC (2010).
In Chapter 5 I also explained that the innovative structure can be approximated by
innovation investments (the intensity and balance of these), capabilities (qualified human
resources and linkages), and the access to external resources. If the strategies are combined
with the characteristics of the structure, the methodological question can be translated into
hypotheses (H) and empirically verified:
Q1: Is there a positive relationship between the continuity of the innovative strategy and
the complexity of the innovative structure?
H1.1: Continuous innovative firms have a more complex innovative structure.
Chapter 6: Regularities and changes in firms’ strategies
- 180 -
H1.2: New innovative firms changed their innovative strategy in the direction of a more
complex structure over time.
H1.3: Sporadic innovative firms have a low-complexity innovative structure sustained over
time.
The hypotheses state that there is a positive relationship between the strategy and the
structure in terms of complexity. H1.1. affirms that firms with a continuous innovative
strategy have a more complex structure, which means that these firms have higher relative
levels of investments in innovation, which are more balanced between activities; higher
capabilities (higher qualified human resources and linkage levels), and more access to
external resources. Of course, the underlying assumption is “the more the better,” although
the average low commitment to innovation among Argentinean manufacturing firms (see
Chapter 2) makes this assumption a plausible one. To measure this complexity, the average
sectoral level for each firm (two-digit ISIC) will be used: firms where each component of
the structure is over that level will be assumed to have a more complex structure.
H1.2 is about firms that changed their strategy and shifted from a non-innovative towards
a continuous innovative strategy. If firms with a continuous innovative strategy have a
more complex innovative structure, then new innovative firms should also present a
change in the innovative structure (from a low to a highly complex one).
Finally, H1.3. is aimed to test the existence of firms where innovation is not the main
strategy. Sporadic firms are supposed to be firms where innovation is an isolated event. If
these firms do not pursue an innovation-based strategy, then their structure should be
similar to that of non-innovative firms, which is expected to have low complexity. In this
respect, although the only variables that could reach positive numbers are qualified human
resources and linkages with the NIS, I will demonstrate that they present significant
differences in respect of the values reached among contiuously and new firms.
6.1.2. The model
Table 6.2 summarizes the variables to be used to analyze the strategies, which is based on
the analyses presented in Chapter 5. Firm has chosen an innovative strategy during
period when it allocated resources to such activities . From the combination
of past and present innovation expenditures the firm will be classified as continuous, new
or sporadic innovative firm ( , respectively) when Iexp>0. If Iexp=0,
the firm will be classified as non-innovative firm. Consequently, its innovative strategy is a
Chapter 6: Regularities and changes in firms’ strategies
- 181 -
function of its past innovative strategy ( ), its innovative structure ( , and its
environmental determinants ( . The innovative structure can be analyzed by looking
at the innovation expenditure in terms of intensity and balance , the endowment
of qualified human resources , interactions with other NIS institutions ,
and the access to external funds ( . The innovation process is expected to be affected
by the impact of the environment and other sources of heterogeneity: firm size ( ,
capital ownership , sectoral belonging ( , and the macroeconomic cycle
(Eqs. 1.1 to 1.4—H1.1 to H1.3).
Formally, the model can be written as follows:
{
(Eq. 1.1)
Where:
(Eq. 1.2)
(Eq. 1.3)
(Eq. 1.4)
Table 6.2: Summary of variables
Label Detail Calculation Value
Innovative Strategy
Innovation expenditure >0 at least one year of the sub-period 0;1
Continuous innovative firm >0 in t & t-1 / t-1 & t-2 0;1
New innovative firm >0 in t / t-1 0;1
Sporadic innovative firm >0 in t, t-1 / t-2 0;1
Non-innovative firm =0 in t & t-1 / t-1 & t-2 0;1
Innovative Structure
Expenditure on innovation
Innovation intensity Innovation expenditure to sales. Annual average
for the sub-period. 0–
Innovation balance Distribution of the total innovation expenditure. 0-
Capabilities
Qualified human resources Highly skilled employees to total employment.
Annual average for the sub-period. 0–
Linkages At least one linkage in the sub-period. Variable
available for t and t-2. 0;1
Resources
Access to external sources of funding At least one source of funding in the sub-period. 0;1
Environmental determinants (structural variables)
Total employment Total employment, average value in t-2. 0–
Capital ownership A firm in which more than 1% of shares are
owned by foreign capital in t-1 0;1
Sectoral belonging 2-digit ISIC classification Rev. 3 in t-1. 0;1
Sub-period One dummy for 0;1
Chapter 6: Regularities and changes in firms’ strategies
- 182 -
6.1.2.1. The descriptive approach
To obtain a first approximation of the strategies, statistical differences among the variables
included in the innovative structure were tested (innovation intensity and balance, qualified
human resources, linkages and access to external funds). The Kruskal–Wallis (1952) and
the Jonckheere–Terpstra (Jonckheere, 1954; Juneau, 2006) tests were performed on the
continuous variables (II, IB, and QHR)83. To include sectoral controls, each of these
variables was transformed into the quotient between the firm’s level and the average level
of the sector it belongs to, based on the 2-digit ISIC classification.
The Kruskal–Wallis (KW) test checks if the groups come from the same population based
on a mean test on the ordered values. The average position of each group is then
compared, based on a chi-square distribution under the null hypothesis that the categories
come from the same population (non-significant differences in means). Based on the
ordered values, the Jonckheere-Terpstra (JT) test analyzes the existence of any order
among the strategies under the null hypothesis of non-significant relationship between the
ordering of the groups and the level of the analyzed variable.
Pearson’s Chi-square and Kendall’s Tau-c tests were performed on the binary and
categorical variables (Link, ER, and strategies). The hypotheses to be contrasted are the
same as with the KW and JT tests, respectively: the probability of a positive response
(dummy=1)and the relationship between an increase in the probability of a positive
response and the order (ranking) of strategies.84
6.1.2.2. The cluster analysis
After the descriptive analysis, a multivariate analysis of variances (MANOVA) and a
confirmatory cluster analysis will be performed. The former will provide evidence about
the existence of underlying matrixes (a specific innovative structure) for each group (each
possible combination of past and present strategies) and the latter will characterize them.
MANOVA is an extension of the traditional ANOVA as it allows multiple dependent
combined variables (a vector) as well as interactive terms among the independent ones.
When the interactive term includes time, MANOVA takes serial correlation among
83 The advantage of both tests is that assumptions of normality are not required so groups’ average values are compared directly. 84 Appendix 2.1 summarizes the main characteristics of the mean tests used for the descriptive statistics.
Chapter 6: Regularities and changes in firms’ strategies
- 183 -
variables into account. Consequently, MANOVA tests whether the vectors of means for
two or more groups come from the same sample distribution, using dynamic estimations.85
Besides the variables associated with the innovative structure (II, IB, QHR, Link and ER),
total labor in t-2 (1998–2001) and capital ownership of the firm (OK) were included in the
estimation.86 As with the descriptive approach, continuous variables were transformed into
the coefficient between the value reached by the firm and the average value for its sector of
belonging at 2-digit ISIC classification to control sectoral specificities.
After MANOVA, a confirmatory cluster technique based on a linear discriminant analysis
will be run in order to identify the levels of each vector of means (if any) and possible
associations between these levels and the independent variables (different structures
associated with different strategies). The discriminant analysis is based on the identification
of discriminant functions (which are the same as the factor analysis scores) which are linear
combinations of variables that separate the cases into specific groups based on the
maximization of distances between categories (Fix and Hodges, 1951; Anderson, 2003).87
As is the case with k-means cluster analysis, the discriminant analysis requires the number
of groups (clusters) to be provided. However, instead of forcing the cluster membership of
cases, this analysis estimates the probability of a correct classification between the provided
cluster membership and the clusters that arise from similar mean vectors. The estimation of
the discriminant functions is formally defined as follows:
Where is the discriminant function for the group ( with group 1 being the
control group) in case (800 cases); the average coefficient of each variable , is the
value of the variable for the case , with the number of predictor variables
85 Standard statistical software estimates four MANOVA statistics: Wilks’ lambda, Pillai’s trace, Lawley–Hotelling trace, and Roy’s largest root. The four statistics are admissible, unbiased, and invariant, and the correlations between covariances of errors should tend to zero (Pillai, 1985). Differences among the statistics arise from assumptions about the distribution of F, so if they are observed together, more robust conclusions can be obtained. A detailed explanation of the MANOVA test and the statistics used in this study is provided in Appendix 2.1.4. 86 Adding more size controls to the estimations (labor in other sub-periods) does not change the final estimations. 87 The discriminant analysis is based on Fisher’s (1936) linear discriminant functions and Mahalanobis’s
(1936) extension to the analysis of group classification. This analysis is based on the identification of the maximum separation between groups given by an eigen analysis between the between-group sum-of-squares and cross-products (SSCP) and the within-group SSCP matrix.
Chapter 6: Regularities and changes in firms’ strategies
- 184 -
. A constant is added to control the different scales and starting values of each
category and variable.88
Based on the discriminant functions, the cases are then assigned to a predicted category
based on the lowest Mahalanobis distance (Mahalanobis, 1936). Formally, the Mahalanobis
distance for the vector of means of the group (with the number of groups equal to the
number of observed categories), estimated for the discriminant functions ( ), from a
panel with mean and covariance matrix S is:
√
Mahalanobis distance is a multivariate technique to test squared distances among vectors.89
Subsequently, estimated and observed classifications are contrasted, together with the
analysis of the incidence of each variable on the category and the probability of a correct or
wrong classification.
6.2. Results from the descriptive analysis
6.2.1. The innovative strategies
Table 6.3 summarizes the transition of the strategies among the sub-periods. On the one
hand, there is a high correlation within the continuous and the non-innovative groups: 88%
of the firms that were continuous during 1998–2004 remain as such during 2002–2006, a
percentage equal to 67% among the non-innovative. At both points in time, the non-
innovative firms constitute a small group (26% and 24% of the panel, respectively) and this
underrepresentation is explained by a high rate of mortality among them, leading to their
disappearance from the statistics (Chapter 5.1.4).
88 As in the factor analysis, there is assumed to be no correlation between the discriminant functions.The mean and variance of the predictors (the variables) are also assumed to be uncorrelated. Likewise, the distribution of predictors is assumed to be normal and cases to be independent. However, discriminant analysis has been proved to be robust even when these assumptions are not fulfilled (Lachenbruch, 1975; Klecka, 1980). 89 Mahalanobis distance is similar to Euclidean distance, although since it is based on the inverse of the covariance matrix, it standardizes the variables to the same variance while eliminating correlations (Rencher, 2002).
Chapter 6: Regularities and changes in firms’ strategies
- 185 -
Table 6.3: Innovative strategies—Number of firms (Pearson’s std. residuals)
Sub-period
Strategies
1998–2001 2002–2004 2005–2006
Innovative Continuous Continuous
446
366 (11.2)
325 (8.5)
Sporadic
41 (2.0)
Sporadic New
80 (5.2)
33 (7.3)
NI
47 (6.5)
Non-Innovative New Continuous
145 (10.0)
120 (4.4)
Sporadic
354 25 (3.8)
NI New
209 (11.8)
67 (8.0)
NI
142 (13.2) Obs.:800. Pearsons’standarized residuals +/-1.96 means significant at 95%. Overall Pearson’s Chi-square significant at 99%. See Appendix 2.2.1. Source: own elaboration based on INDEC (2010).
The group of continuous innovative firms is made up of firms that, regardless of the macro
context, have sustained their innovation investments over time. This group represents 46%
and 56% of the panel in 2002–2004 and 2005-2006, respectively. Of course, given the poor
innovative performance of Argentinean firms (Chapter 2), this high participation is
probably explained by the bias in the sample and not by the actual participation of high-
profile innovative firms within the manufacturing sector. The high correlation between the
continuous groups in each sub-period could be a manifestation of the exit costs associated
with investments in innovation as well as in possible long-term projects, only captured
when expenditures are considered in the long run (Chapter 3.3.2.1). Regarding non-
innovative firm, their existence proves that innovation is one of the possible strategies
firms can choose, there are many other competitive advantages which can also allow the
firm to survive and grow (Chapter 3.1.2).
The transition of the newly innovative firms to a continuous strategy is also important:
more than 80% of those firms which started investing in innovation during 2002–2004
continued with investments during 2005–2006. Overall, 15% of the panel changed their
strategy: from null investments in innovation between 1998 and 2001 to a sustained
investment between 2002 and 2006. These are the firms that would be changing their
innovative strategy and not just undertaking an opportunistic investment.
Chapter 6: Regularities and changes in firms’ strategies
- 186 -
Finally, the sporadic innovative group represents 10% and 8% of the panel. Sporadic
investments are usually found in firms which are not systematic seeking technological
and/or organizational improvements, where innovation is associated with the
implementation of specific short-term improvements (mostly due to the acquisition of
capital goods) (see Chapter 4.1). The evolution of the sporadic group is as expected: 60%
decided not to make further investments and shifted from sporadic to non-innovative; the
other 40% make investments during 2005–2006 but not continuously. In 2005–2006, the
group of sporadic firms was joined by 25 firms that made investments right after the
change in the macro scenario (2002–2004).
Another interesting aspect of the different strategies is the actual frequency of expenditure
(Table 6.4). Of the nine years considered, continuous firms invested in innovation in
almost eight, with a higher frequency than the rest of the strategies, even for the period
1998–2001. And although new innovative firms did not invest in innovation during this
first sub-period (frequency=0), they catch up with the frequency of continuous firms by
2006 (frequency=4.1 between 2002–2004 versus 4.4 among continuous firms). Of course,
among sporadic and non-innovative firms the frequency decreases and the differences with
the rest of the strategies are remarkable throughout the period.
Table 6.4: Frequency of expenditure—Number of years—Mean (Std. dv.)
Sub-period
Strategies
1998–2001 2002–2004 2005–2006
Innovative Continuous Continuous*
3.4 (0.97) 2.5 (0.79) 1.8 (0.37)
Sporadic
3.0 (1.16) 1.6 (0.89) 0.0 (0.0)
Sporadic New
2.7 (1.24) 0.0 (0.0) 1.4 (0.56)
NI
2.6 (1.25) 0.0 (0.0) 0.0 (0.0)
Non-Innovative
New Continuous*
0.0 (0.0) 2.3 (0.84) 1.8 (0.40)
Sporadic
0.0 (0.0) 1.9 (0.81) 0.0 (0.0)
NI New
0.0 (0.0) 0.0 (0.0) 1.4 (0.53)
NI
0.0 (0.0) 0.0 (0.0) 0.0 (0.0) Obs.: 800. KW and JT positive and significant at 99% for the three sub-periods and the eight combined strategies, except from the pairs marked with *. Complete estimations in Appendix 2.2.1. Source: own elaboration based on INDEC (2010).
Chapter 6: Regularities and changes in firms’ strategies
- 187 -
This table constitutes a first approximation to the predictive power of the classification and
the importance of analyzing firms in terms of an evolving strategy. As I shall show in the
following section, the firm’s innovative structure can be predicted by looking at the firm’s
innovative strategy between the first two sub-periods and the classification for the
following one, based on which predictions can be made regarding the trajectory of the
innovation expenditure, its distribution among activities, the increase or decrease in the
level of qualified human resources, and the interaction with the rest of the NIS. In the case
of the access to external resources, the relationship is not so clear, although it coincides
with investment patterns among Argentinean firms (Chapter 2.3).
6.2.2. The strategies and the innovative structure
This section aims to describe the relationship between the innovative strategies and the
innovative structures. Innovation expenditure and balance—the main indicators of the
structure—evolve along very different trajectories in each of the different innovative
strategies (Table 6.5).
Table 6.5: Innovation intensity (Iexp) and balance (IB)–Mean (Std. dv.)
Sub-period Sub-period
Iexp
1998–2001 2002–2004 2005–2006
IB
1998–2001 2002–2004 2005–2006
Innovative ContinuousX ContinuousXX Innovative Continuous Continuous
3.54 (6.39) 4.24 (5.76) 2.25 (2.79) 0.43 (0.21) 0.38 (0.22) 0.38 (0.21)
Sporadic Sporadic
2.64 (4.19) 2.68 (3.78) 0 (0.0) 0.35 (0.22) 0.19 (0.17) 0
Sporadic New Sporadic New
3.27 (5.18) 0 (0.0) 1.86 (3.09) 0.29 (0.21) 0 0.31 (0.20)
NI NI
1.09 (1.52) 0 (0.0) 0 (0.0) 0.29 (0.18) 0 0
Non-Innovative
NewX ContinuousXX Non-
innovative New Continuous
0 (0.0) 5.04 (9.50) 2.46 (3.96) 0 (0.0) 0.33 (0.21) 0.34 (0.20)
Sporadic Sporadic
0 (0.0) 5.02 (7.42) 0 (0.0) 0 (0.0) 0.18 (0.14) 0 (0.0)
NI New NI New
0 (0.0) 0 (0.0) 1.40 (2.62) 0 (0.0) 0 (0.0) 0.24 (0.18)
NI NI
0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) Obs.: 800. Iexp: innovation expenditure to sales, average values for each sub-period. IB: innovation balance, average values for each sub-period. KW and JT positive and significant at 99%, except for the pairs marked
with X and XX. Complete estimations in Appendix 2.2.2 and 2.2.3. Source: own elaboration based on INDEC (2010).
With regard to innovation intensity, continuous firms show high relative levels of
expenditure (to sales), which are also sustained throughout all the periods analyzed (3.54%,
Chapter 6: Regularities and changes in firms’ strategies
- 188 -
4.24%, and 2.25%, respectively). Despite the drop in 2005–2006, the levels are still above
the panel average and they are only surpassed by those of new innovative firms during
2002–2004. Among the continuous firms there are, however, a group of firms that change
their innovative strategy: the C-SP. These firms invested in innovation during the crisis and
the recovery period (1998–2001 and 2002–2004) but stopped doing so subsequently.
Among these firms, the relative investments in innovation are low throughout the entire
period analyzed.
Three different trajectories can be observed when it comes to the group of firms that
started investing in innovation after the change in the macroeconomic scenario, the new
innovative firms. First, there is a group of firms that started investing during 2002–2004
and continued with high levels of investments during the following sub-period (N-C), with
similar levels to those of the C-C firms. These results could be a sign of a real change in the
innovative strategy, with a proportionally higher level at the beginning (the initial sunk
costs) and a level similar to that of more innovative firms afterwards. Secondly, there is a
group of firms with a high level of expenditure during 2002–2004 but zero investments in
any other period (N-S). This group is probably the result of short-term strategies that
aimed to take advantage of the exchange rate and the boost in domestic demand, but
without a long-term strategy. Although the group’s level of expenditure is significantly high
(5.02), its investment behavior is similar to that observed among the sporadic firms, that is,
isolated innovation investments. Of course, although short-term positive impacts are
expected, the hypothesis among these firms is that their strategies will fall short of leading
them to sustainable growth paths.
The distribution of the expenditure (IB) on innovation fulfills what the theory predicts and
is consistent with the patterns of intensity. There seems to be a high correlation between
the continuity of the expenditure and higher levels of balance. Both the C-C and N-C firms
show similar IB values. These levels are the results of combined investments in
endogenous and exogenous knowledge sources (capital goods, R&D, training, engineering)
sustained over time. The rest of the strategies, the sporadic firms in particular, present
lower values, which are explained by a bias in the expenditure towards the acquisition of
capital goods with a generalized lacked of investments in complementary activities (Table
6.6).
Chapter 6: Regularities and changes in firms’ strategies
- 189 -
Table 6.6: Distribution of the innovation expenditure
C-C C-SP SP-N SP-NI N-C N-SP NI-N
t-2 t-1 t t-2 t-1 t t-2 t-1 t t-2 t-1 t t-2 t-1 t t-2 t-1 t t-2 t-1 t
R&D (internal + external) 0.20 0.25 0.27 0.11 0.13
0.2
0.13 0.25
0.24 0.19
0.1
0.13
Engineering and industrial design + training
0.15 0.11 0.12 0.28 0.08
0.1
0.1 0.18
0.17 0.14
0.13
0.1
Capital goods + Hardware 0.48 0.57 0.5 0.46 0.78
0.48
0.56 0.34
0.48 0.56
0.76
0.59
Technology transfer + Consulting + Software
0.17 0.07 0.12 0.15 0.01
0.21
0.22 0.23
0.1 0.11
0.01
0.18
Total Iexp 1.00 1.00 1.01 1.00 1.00
0.99
1.01 1.00
0.99 1.00
1.00
1.00
Obs.: 800. Iexp: innovation expenditure to sales. NI-NI omitted. t-2: 1998–2001; t-1: 2002–2004; t: 2005–2006. Source: own elaboration based on INDEC (2010).
The analysis of qualified human resources and linkages (capabilities) sheds light on
differential features of the strategies and confirms the predictive power of the classification:
high levels of skilled personnel and higher probabilities of linkages are associated with
sustained investments in innovation while low levels are associated with erratic or non-
innovative behaviors (Table 6.7). The C-C and the NI-NI firms show the expected results,
the former with the highest levels of qualified human resources (11.33% of total
employment), the latter with the lowest values (4.84%). In between, two different trends
are observed.
Table 6.7: Qualified human resources (QHR) and linkages (Link)
Sub-period—Mean (Std. dv.) Sub-period—% firms (Std. residuals)
QHR
1998–2001 2002–2004 2005–2006
Link
1998–2001 2002–2004 2005–2006
Innovative Continuous Continuous Innovative Continuous Continuous
10.11 (11.78) 8.64 (12.53) 11.33 (15.00) 85.23* (5.1) 69.54* (5.7)
Sporadic Sporadic
4.65 (5.28) 3.67 (4.50) 6.38 (8.05) 78.05 (1.2) 14.63* (-3.1)
Sporadic New Sporadic New
7.03 (8.36) 7.83 (9.10) 11.63 (12.01) 81.82 (1.4) 42.42 (-0.4)
NI NI
8.56 (14.72) 4.85 (9.27) 10.23 (14.13) 76.60 (1.2) 12.77* (-3.5)
Non-Innovative
New Continuous Non-
Innovative New Continuous
6.85 (12.30) 7.50 (10.83) 9.90 (13.17) 52.50 (-1.4) 65.00* (2.8)
Sporadic Sporadic
2.41 (4.52) 3.64 (6.52) 6.02 (6.65) 16.00* (-3.0) 20.00* (-2.0)
NI New NI New
6.20 (11.09) 4.72 (9.03) 7.31 (9.80) 35.82* (-2.8) 43.28 (-0.5)
NI NI
2.58 (5.56) 2.84 (5.20) 4.84 (10.29) 28.17* (-5.2) 11.97* (-6.2) Obs.: 800. QHR: skilled employees to total employment, average values for each sub-period. KW and JT positive and significant at 99% for t-2 and 95% for t-1 and t for the eight combined strategies. Links: at least one link with other NIS organizations, percentage of firms. *: significant at 99%, sign given by the sign of the std. residuals (Pearson’s Chi-square and Kendall’s Tau-c tests). Complete estimations in Appendix 2.2.4 and 2.2.5. Source: own elaboration based on INDEC (2010).
Chapter 6: Regularities and changes in firms’ strategies
- 190 -
On the one hand, sporadic innovative firms present low levels of qualified human
resources, although they surpass those of non-innovative firms. On the other hand, the
results of new innovative firms that continued to invest in innovation (N-C) exceed those
of the rest of the firms due to the significant increase in this indicator. A catching-up
process is observed in skills that is similar to the supposed catching-up process associated
with expenditure, with sustained increases in the level of qualified employment.
With regard to firms’ probability of having interacted with any NIS organization, once
again, the highest levels are observed among the continuous firms and the lowest among
the non-innovative ones. N-C firms stand out again, with high probabilities of interaction
even before starting to invest in innovation (1998–2001).
A characteristic shared by firms with high levels of interactions (C-C and N-C) is the
presence of multiple linkages. Indeed, firms reported that they interacted with S&T
institutions and suppliers and customers as part of their innovation activities (Table 6.8).
These results confirm what is predicted by the NIS when innovation is assumed to be a
systemic process and also contribute to the confirmation of a change in innovative strategy:
once again N-C firms tend to resemble the behavior of C-C ones.
Table 6.8: Linkages with organizations from the NIS
C-C C-SP SP-N SP-NI N-C N-SP NI-N NI-NI
t-2 t t-2 t t-2 t t-2 t t-2 t t-2 t t-2 t t-2 t
S&T 72.3 49.2 48.8 12.2 48.5 6.1 51.1 6.4 37.5 46.7 8 8 22.4 25.4 14.1 5.6
Gov. 23.7 11.4 14.6 2.4 3.03 6.1 23.4 0 12.5 15 4 0 7.5 1.5 2.1 1.4
Cli. 70.8 31.7 56.1 0 69.7 24.2 68.1 8.5 39.2 35.8 12 16 25.4 13.4 18.3 4.9
Sup. 64.3 46.1 51.2 7.3 54.5 33.3 61.7 6.4 32.5 45.8 12 12 20.9 22.4 14.8 6.3
Obs.: 800. S&T: University + Institute of Technical Training + Technological center+consultants+R&D Labs; Gov: Agencies and public programs of S&T promotion; Cli: Clients; Sup: Suppliers. Source: own elaboration based on INDEC (2010).
Table 6.9 shows the last innovative variable included: access to external funds. In this case,
the expected correlation between the level of investment and the probability of having
accessed external sources of funding is observed. One interesting finding, however, is the
fact that among the C-C and N-C firms (the ones with the highest levels in all indicators)
the variability of ER is higher than the variability of the level of investments. Conversely,
among sporadic firms, dips and peaks in the evolution of the two indicators (innovation
expenditures and external resources) happen at the same. This differentiated co-evolution
would reinforce the idea of different innovative strategies, one that is more complex among
the continuous and new innovative firms and one that is more opportunistic among
Chapter 6: Regularities and changes in firms’ strategies
- 191 -
sporadic firms, which arises from the possibility of accessing external funding (specifically,
the acquisition of capital goods with payment facilities offered by the suppliers90).
Table 6.9: Access to external resources (ER)—% of firms (std. residuals)
Sub-period
ER
1998–2001 2002–2004 2005–2006
Innovative Continuous Continuous
35.08* (6.3) 11.69* (2.1) 22.46* (4.8)
Sporadic
31.71 (1.7) 7.32 (0.2) 0
Sporadic New
39.39* (2.6) 0 21.21 (1.3)
NI
34.04* (2.2) 0 0
Non-innovative New Continuous
0 15.83* (2.8) 14.17 (0.4)
Sporadic
0 20.00* (2.0) 0
NI New
0 0 8.96 (-0.9)
NI
0 0 0 Obs.: 800. ER: access to at least one external source of funds, percentage of firms. *: significant at 99%, sign given by the sign of the std. residuals (Pearson’s Chi-square and Kendall’s Tau-c tests). See Appendix 2.2.6. Source: own elaboration based on INDEC (2010).
6.3. Results from the cluster analysis
Given the evidence on similar trends between innovative variables and strategies, this
section aims to test for the existence of a significant association between strategies and the
different innovative structures (H1.1 to H1.3).
Table 6.10 presents the results of the MANOVA and provides preliminary confirmation of
the existence of different vectors of means for the different strategies. The four MANOVA
statistics are significant, meaning that specific combinations of the variables are associated
with the innovative structure (there is a specific vector for each combination of strategies).
It also shows that both time and strategies are significant classification criteria, and this
finding supports the hypothesis about the joint impact of the environment and the
innovative strategy. In other words, the identified strategies are valid classification variables
when time is controlled.
90 This was, in fact, the generalized behavior during the nineties.
Chapter 6: Regularities and changes in firms’ strategies
- 192 -
Table 6.10: MANOVA
Model
Wilks’ lambda 0.3512* 0.4689* 0.9788* 0.8355*
Pillai’s trace 0.78* 0.5721* 0.0212* 0.1664*
Lawley-Hotelling trace 1.4943* 1.0469* 0.0217* 0.1946*
Roy’s largest root 1.2367* 0.9608* 0.0217* 0.1823* Obs.: 2400. *Significant at 99%. Residuals: 1599. t-2: control group. Marginal effects positive and significant.
( ) Complete estimations in Appendix 2.3. Source:
own elaboration based on INDEC (2010).
Using the same variables that account for the innovative structure —each of them for each
sub-period— plus the size, sector, and capital ownership controls, seven discriminant
functions were estimated. Table 6.11 summarizes them and average value for each variable.
Table 6.11: Canonical structure (coefficients of the discriminant functions)
Discriminant functions
1 2 3 4 5 6 7
t-2 0.216 -0.102 -0.045 -0.063 0.176 0.503 -0.555
t-1 0.092 0.226 -0.349 0.377 0.477 -0.002 -0.314
T 0.108 0.204 0.152 0.056 -0.052 -0.057 -0.172
t-2 0.749 -0.389 -0.034 -0.099 0.140 -0.217 0.057
t-1 0.445 0.530 -0.511 -0.071 -0.219 -0.044 0.167
T 0.403 0.542 0.463 -0.045 0.148 0.124 0.063
t-2 0.132 0.010 0.191 0.232 -0.024 -0.373 -0.197
t-1 0.023 0.072 0.055 0.359 0.090 0.319 0.277
T 0.065 0.004 0.167 0.418 0.251 -0.176 0.336
t-2 0.360 -0.068 0.070 0.619 -0.544 0.094 -0.050
T 0.267 0.396 0.185 -0.089 -0.046 -0.302 -0.027
t-2 0.280 -0.198 0.047 0.252 0.160 -0.023 0.267
t-1 0.072 0.170 -0.165 0.195 0.430 -0.110 0.068
T 0.162 0.138 0.160 -0.097 0.212 0.194 0.261
t-1 0.118 0.064 -0.060 -0.092 0.022 0.063 0.181
t-2 0.157 0.059 -0.008 -0.276 -0.157 -0.258 0.250
Eigenvalue 2.458 0.914 0.329 0.045 0.017 0.010 0.008
Accumulated % 65.0 89.2 97.9 99.1 99.5 99.8 100.0
Canonical correlation 0.843* 0.691* 0.498* 0.207 0.128 0.099 0.088
Obs.: 800. *: significant at 99%. Canonical correlation based on Wilks’ lambda contrast of the function, significant at 99.9%. Joint within-group correlations between discriminant variables and standardized canonical functions. Source: own elaboration based on INDEC (2010).
Functions 1 to 3 (out of seven) present a significant canonical correlation (an underlying
vector of means) and explain 98% of the variability among strategies over time, Functions
1 and 2 with eigenvalues closer than one and a canonical correlation over 0.7. Function 1
accounts for those firms with a high-profile innovative strategy during the first sub-period:
high levels of balanced innovation investments, qualified human resources, high
probabilities of interacting with the NIS and accessing external resources (in all cases, for
1998–2001). Function 2 accounts for the same situation but for the following two sub-
Chapter 6: Regularities and changes in firms’ strategies
- 193 -
periods (2002–2004 and 2005–2006) in the case of the innovation investments (intensity
and balance), and for the second sub-period (2002–2004) in the case of human resources,
linkages, and access to external funds. According to the descriptive statistics, continuous
firms should rank high in both of these functions while new innovative firms should only
do so in the second one.
Although Function 3 has a very marginal impact on the variability among cases, it
represents those firms with high levels of human resources, combined with lower levels of
expenditure, balance, and linkages. In this latter case, access to external resources is high
for the last sub-period, together with an increase in the level of investments, which is most
probably associated with the behavior of sporadic firms, according to what was observed in
the descriptive statistics.
Graph 6.1 illustrates the cluster estimation based on the significant discriminant functions
(1 to 98% of the sample’s variance), while Graphs 6.2.a and 6.2.b represents the same
function but on two axes. Table 6.12 summarizes the mean and the standard deviation of
the significant discriminant functions for each strategy.
Graph 6.1: Strategies and discriminant functions (DF)
Obs.: 800. Non-standardized canonical discriminant functions, mean values. See Appendix 2.4 for descriptive statistics and tests that confirm the significant differences. Source: own elaboration based on INDEC (2010).
As we can observe in Graph 6.2.a, C-C firms rank high in Functions 1 and 2, which means
that their innovative structure is characterized by high and balanced levels of innovation
DF 3
DF 2
DF 1
-1 1
1
1
-1
Chapter 6: Regularities and changes in firms’ strategies
- 194 -
expenditures, high levels of qualified human resources, and linkages sustained over time. C-
N firms also rank high in Function 2, but negatively in function 1, which means that, on
average, they show a similar innovative structure to C-C firms but starting from t-1. As
expected, NI-NI firms present the opposite situation. These firms rank below zero in both
functions, meaning they have low levels of investments, qualified human resources,
linkages, and low dispersion, with levels close to zero. The rest of the strategies show a
high dispersion in terms of Function 1 but they tend to cluster together around closer
negative values in Function 2. The high dispersion in Function 1 means that these firms
present different combinations of investments, capabilities, and resources, although they
are not sustained over time—indeed, they are specifically concentrated around t-2.
Graph 6.2.a: Discriminant Function 1 versus Discriminant Function 2
Obs.: 800. Non-standardized canonical discriminant functions . Source: own elaboration based on INDEC (2010).
If Function 3 is added to the analysis (Graph 6.2.b), the lower explicative power of this
function is confirmed, with all firms concentrated on values around zero. The lower impact
of the level of qualified human resources on explaining the different strategies is also
confirmed, once the balance, intensity, and linkages have been controlled. The most
interesting finding for this function is the higher values registered among N-C, which
means that these firms present high levels of capabilities before changing their strategy, an
aspect already observed in the descriptive statistics. Another interesting element of this
function are the higher values observed among new innovative firms in the last sub-period
(NI-N and SP-N), associated with the increase in the qualified human resources in t. This
behavior resembles the origin of N-C two sub-periods earlier: the increase in capabilities
-1
0
1
-1 0 1
NI-NI NI-N
SP-NI SP-N
N-SP N-C
C-SP C-C
DF 1
DF 2
Chapter 6: Regularities and changes in firms’ strategies
- 195 -
takes place before the increase in the level of investments. The preliminary conclusion
about this is that evidence exists to support the hypothesis of a new set of new innovative
firms starting from 2005, a hypothesis that, of course, can only be answered by extending
the period under analysis.
Graph 6.2.b: Discriminant Function 1 vs. Discriminant Function 3
Obs.: 800. Non-standardized canonical discriminant functions. Source: own elaboration based on INDEC (2010).
Table 6.12: Strategies and discriminant functions–Mean values (std. dv.)
Discriminant functions
1 2 3
NINI -2.106 (0.362) -0.601 (0.263) -0.105 (0.319)
NIN -1.686 (0.468) 0.267 (0.544) 1.042 (0.721)
SPNI -0.339 (0.826) -1.913 (0.725) 0.060 (0.428)
SPN 0.440 (1.239) -0.975 (0.815) 1.479 (0.811)
NSP -1.893 (0.486) 0.303 (0.768) -1.461 (1.098)
NC -0.874 (0.677) 1.747 (1.234) -0.173 (1.403)
CSP 0.410 (1.179) -1.499 (0.905) -1.182 (0.881)
CC 1.689 (1.318) 0.104 (1.229) -0.003 (1.138) Obs.: 800. Non-standardized canonical discriminant functions. Average value for each group. Source: own elaboration based on INDEC (2010).
Table 6.13 shows final cluster estimations. The classification criterion of the strategies (the
frequency) accounts for significant differences in the innovative structure (investments,
capabilities, and resources) and the cluster membership based on these variables explains
77.5% of the firms. That is to say, when firms are classified according to their innovative
strategy over time, more than 3 out of 4 of them will end up within a group with similar
characteristics in terms of innovative structure. In terms of this dissertation, the continuity
of the strategy is positively associated with the complexity of the structure: firms with a
-1
0
1
-1 0 1
NI-NI NI-N
SP-NI SP-N
N-SP N-C
C-SP C-C
DF 1
DF 3
Chapter 6: Regularities and changes in firms’ strategies
- 196 -
continuous innovative strategy have a more sustained complex structure (H1.1) and new
innovative firms shift from a low to a high complex structure (H1.2).
The largest absolute differences between the observed and predicted classifications are
among the firms predicted to be non-innovative, where the estimation places also many
sporadic innovative firms. This finding is consistent with H1.3: those firms with a sporadic
innovative strategy present a similar innovative structure to the non-innovative firms.
Table 6.13: Cluster analysis - Estimated and observed cluster membership
Observed Estimated
Total NINI NIN SPNI SPN NSP NC CSP CC
% of cases
NINI 97.9 1.4 .7 0.0 0.0 0.0 0.0 0.0 100.0
NIN 37.3 62.7 0.0 0.0 0.0 0.0 0.0 0.0 100.0
SPNI 38.3 0.0 61.7 0.0 0.0 0.0 0.0 0.0 100.0
SPN 18.2 12.1 12.1 33.3 0.0 0.0 0.0 24.2 100.0
NSP 64.0 0.0 0.0 0.0 24.0 12.0 0.0 0.0 100.0
NC 14.2 8.3 0.0 0.0 1.7 75.8 0.0 0.0 100.0
CSP 24.4 0.0 9.8 0.0 2.4 0.0 53.7 9.8 100.0
CC 2.5 .9 1.2 1.2 0.0 7.4 .6 86.2 100.0
Error rate 0.02 0.37 0.38 0.70 0.76 0.24 0.46 0.14 0.225
Obs.: 800. 77.5% of cases correctly classified, estimations based on within-groups covariance. Error rate based on error count. 93% correctly classified if the separated groups covariance matrix is used. Wilks’ lambda significant at 99% for all variables. Mahalanobis distances significant for all groups. See Appendix 2.4. Source: own elaboration based on INDEC (2010).
6.4. Discussion of the findings
Graph 6.3 summarizes the main findings of the empirical analysis based on the average
values (Y axis) of the innovative structure (X axis) for each sub-period, for the main
strategies. During 1998–2006, four types of innovative behaviors can be observed among
Argentinean manufacturing firms: continuous (C-C), new (N-C), non-innovative (NI-NI),
and sporadic firms (remaining categories).
The results show that the four different strategies group firms with similar innovative
structures: continuous firms present a high level of complementary investments and
capabilities; new firms seem to have caught up in terms of innovation expenditures based
on pre-existent high levels of capabilities; and sporadic firms show opportunistic behavior,
based on short-term process innovations associated with the incorporation of machinery or
small investments aimed at solving specific problems with low impacts on the innovative
dynamic. Of course, non-innovative firms show the poorest results and a consistent non-
Chapter 6: Regularities and changes in firms’ strategies
- 197 -
innovative structure over time (low levels of investments, qualified human resources, and
linkages).
Graph 6.3: Innovative strategies and innovative structure
Obs.: 800. Continuous: C-C; New: N-C; Non-innovative: NI-NI; Sporadic: remaining firms. Values estimated as the quotient between the average values for the strategy and the panel, for each sub-period. Source: own elaboration based on INDEC (2010).
According to these results, the answer to Q1 is yes, there is a positive relationship between
the continuity of the innovative strategy and the complexity of the innovative structure.
Therefore, the continuity of the innovation expenditure is a good criterion for the
identification of innovative strategies. Different frequencies of expenditure on innovation
correspond with differences in the innovative structure among groups of firms. Eventually,
those firms with sustained but low levels of investments (as we hypothesized before, firms
that only spend a small amount on simple training or basic activities) become sporadic or
non-innovative firms when they are observed over a long time period.
In terms of changes in the strategy, results show that only new innovative firms do changes
in the strategy correspond with changes in the structure. Among sporadic firms, the
structure remains the same although the frequency of the expenditure changes. These firms
show erratic levels of investment in innovation which seem to be associated with different
probabilities of accessing external funds. In this case, even though these firms eventually
invest in innovation, their structure is close to the structure of non-innovative firms, which
in fact show stable behavior over time.
Regarding the structural determinants, the distribution of the strategies by sector does not
show any significant correlation, which means that the four strategies identified are present
in all groups, and all groups are made up of firms using the four strategies (Graph 6.4). On
the one hand, these findings confirm the possibility of a high-profile innovative strategy in
low-tech sectors (such as textiles) but also, and conversely, the existence of firm with a
0.0
0.5
1.0
1.5
2.0
2.5
1998-2001 2002-2004 2005-2006
II IB QHR Link Rec_ext II IB QHR Link Rec_ext II IB QHR Rec_ext
Contuous SporadicNew Non-Innovative
Chapter 6: Regularities and changes in firms’ strategies
- 198 -
non-innovative strategy in sectors where a high innovative profile is to be expected (such as
chemicals). On the other hand, these findings contradict the literature that finds strong
sectoral correlation when it comes to the innovative behavior (de Jong and Marsili, 2006;
Leiponen and Drejer, 2007). Of course, they agree with the evidence regarding the
possibility of increasing profitability based on the incorporation of knowledge regardless of
the sectoral source of technology (Arvanitis and Hollenstein, 1998; Fraga et al., 2008;
Bottazzi et al., 2010; Srholec and Verspagen, 2012; Yurtseven and Tandoğan, 2012).
Graph 6.4: Composition of the strategies—Selected sectors (% of firms)
Obs.: 800. Sector (ISIC): food and beverage (15), textile, leather and wearing apparel (17–19), paper and publishing (21–, 22), chemical and plastics (24–25), non-metallic minerals (26), metalworking (27–29), electrical and electronic machinery (31, 33), vehicles (34), furniture (36). Complete cross tabulations in Appendix 2.4.7. Source: own elaboration based on INDEC (2010).
Regarding the hypotheses of this chapter, they are all confirmed although additional
clarifications have to be made. Continuous firms present a more complex innovative
structure (H1.1), which means higher levels for all indicators, and even though they are not
the highest in all periods, they are sustained over time (that is, they show low oscillations).
Among the new innovative firms (N-C), the correlation between the initial status and the
evolution of the indicators is quite the opposite. These firms changed from a low- to a
high-complexity innovative structure (H1.2). These firms show a deep change in their
strategy during 2002–2004, when they began to make innovation investments, whereas
there was a total lack in investments in the previous sub-period. However, their capabilities
were present before that. Although this category of firm shares the lack of investment
during 1998–2001 with many non-innovative firms of that period, they present high levels
of qualified human resources and linkages, which could be understood as the platform
from which the innovative strategy was drawn. In this respect, their structure was not
similar to the non-innovative firms to the extent that capabilities were high even when
investments in innovation were non-existent.
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Cont New Spor NI
Food and beverage
Textile, leather and wearing apparel
Paper and publishing
Chemical and plastics
Non-metallic minerals
Metalworking
Electrical and electronic machinery
Vehicles
Furniture
Rest
Chapter 6: Regularities and changes in firms’ strategies
- 199 -
Finally, H1.3 is confirmed. Sporadic innovative firms show isolated investments in
innovation without accumulation of capabilities. Although high investments are observed
within specific sub-periods, they only temporarily modify the innovative structure in terms
of the level of investments. Therefore, firms with a sporadic innovative strategy show a low
complexity innovative structure, similar to the one presented by non-innovative firms.
These results only partially agree with the literature on innovative strategies and the
existence of micro-heterogeneity. They do, however, coincide in terms of the existence of
different innovative strategies that are both the result of different choices made by the firm
and the cause of different innovative structures (Srholec and Verspagen, 2012; Leiponen
and Drejer, 2007; Arvanitis and Hollenstein, 1998; among others).
The differences between this study and the reviewed literature lie in the way national and
sectoral specificities are taken into account. Clausen et al. (2011) and Srholec and
Verspagen (2012) find different innovative strategies, which resemble Pavitt’s taxonomies
but at the firm level. Within their findings, more dynamic innovative firms are those that
have high investment intensity and focus on specific results (product-, market-, process-
oriented, etc.), which are associated with a specific type of expenditure. In this respect, our
results show that high-profile innovative firms are in fact those that balance their
innovation expenditure. As a consequence, the firms the literature interprets as high-profile
innovative firms are biased firms in our study. The explanation of these different results
lies, to some extent, in how sectoral belonging is controlled in the studies reviewed. As
sectoral specificities are not properly accounted for, the results are biased and more
determined by the productive structure than by individual decisions. As such, national
specificities will not be relevant because they are captured by the type of the innovative
structure: there will be more R&D–intensive firms in countries with a higher participation
of high-tech industries (such as Germany), and more technology–buyer firms in those
countries with more participation of traditional large-scale labor intensive sectors (such as
Italy). These types of classifications, although useful for describing general trends between
countries, fall short when accounting for elements through which to identify firms that lie
outside the average levels, which are in fact, the true source of micro-heterogeneity.
Another part of the explanation lies in the static nature of their conceptual approach. From
a static perspective, a firm could focus its investments on one type of activity (such as the
development of a new product). However, it is difficult to think of a firm that, once it has
achieved an innovation, automatically initiates a new innovation process, aimed at reaching
Chapter 6: Regularities and changes in firms’ strategies
- 200 -
a similar result (continuing with the above example, by seeking another new product).
Conversely, once the innovation has been achieved, a different set of investments should
be allocated to exploit and improve it. In contrast, since the literature assumes that there is
a linear association between the type of investments and the results (for example,
expenditure on capital goods leads process innovations), they neglect the impact on
capabilities and the possibility of positive feedbacks from the innovation process into other
areas of the firm are excluded from empirical estimation. The acquisition of capital goods
and investment in engineering could lead to new processes but also to new products. New
products can lead to new commercial channels but also to improving the productive
process, given the capabilities created during the development of this innovation.
Then, existent literature can explain different structures from a static perspective but it
does not adequately account for how the particular set of structures will evolve over time.
The analyses performed in the literature force the firm to focus a single type of structure as
if innovations were an isolated, mono-causal event. Our results show that allowing the firm
to present different innovative structures could contribute to understanding the evolving
nature of the firm’s behavior, how they react to environmental changes, and, especially,
how feedback, accumulation, and learning processes can alter the firm’s innovative
behavior.
In short, results show that examining changes in the levels of the indicators and how the
variables are combined provides evidence regarding the importance of accounting for the
learning processes and the adaptive reaction of firms, that is to say, the dynamic process of
innovation. Snapshot analyses are not enough to capture this phenomenon and could lead
to incorrect conclusions showing low levels that are in fact the result of infinite variations
(from zero to something) and high levels which are part of an isolated short-term process,
so they are prone to decline. From a practical perspective, our results provide evidence
regarding the predictive power of the classification according to the continuity of
innovative expenditures.
6.5. Preliminary conclusions
This chapter aimed to test the existence of different innovative strategies based on the
frequency of innovative expenditure, under the hypothesis of a positive relationship
between a theoretically defined strategy and the innovative structure. In order to do so, a
Chapter 6: Regularities and changes in firms’ strategies
- 201 -
multivariate model was constructed and a confirmatory cluster analysis was run on a panel
made up of 800 manufacturing Argentinean firms with information for 1998–2006.
Results show that some firms persistently invested in innovation, combining endogenous
with exogenous investments in knowledge, with a high level of qualified human resources
and strong interactions with the NIS. Although these firms were affected by Argentina’s
economic crisis (1998–2001), they managed to sustain high levels of investment and a high
endowment of skilled employees. When economic growth started again (2002–2004), they
reinforced their strategy and investments and qualified employment grew.
The composition of this group is heterogeneous in terms of sectors, sizes, and capital
ownership. Therefore, it provides evidence about the possibility of a dynamic (high-profile)
innovative strategy regardless of the structural determinants and the instability of the macro
environment. Moreover, they provide evidence about the need for new classification
criteria when it comes to understanding firm-specificities: comparing innovative with non-
innovative firms is not enough when it comes to the innovation process.
Another group of firms present a shift in the innovative strategy. New innovative firms
decided to start investing in innovation after the crisis and they persisted with that behavior
during the recovery and growth phases. Evidence shows a significant and sustained
increase in the level of expenditures and qualified human resources. There is also a catch-
up process at the beginning of the new strategy that differentiates these firms from the
remaining innovative ones (the different sporadic behaviors) and provides evidence
regarding how deep this change was (the lock-in effect of the selected innovative project).
The “why” of the change in these firms’ strategy of these firms is likely to deserve a
complete research project (including a qualitative approximation), but the evidence
presented in this chapter sheds some light on the subject. Even though these firms shared
their lack of investments in innovation with a larger group of firms at the beginning of the
period, the new innovative firms had pre-existing high levels of human resources and more
linkages with other NIS organizations.91
Another feature of the new innovative firms is that they differ in terms of their sectors,
sizes, and capital ownership. As with the continuous firms, dynamic innovative behavior is
not, as a strategy, exclusive to high-tech firms. Moreover, the balanced expenditure of new 91 Of course, one possibility is that the new innovative firms had started investing in innovation before 1998.
However, theoretically speaking, a firm that does not invest in innovation for such a long period (four years) is probably basing its competitive strategy on another factor.
Chapter 6: Regularities and changes in firms’ strategies
- 202 -
innovative firms provides evidence regarding the importance of both the endogenous and
exogenous creation of knowledge, regardless of the sector’s technological intensity.
The innovative dynamic of sporadic firms is the most heterogeneous, although all sporadic
firms share erratic behavior. The level of expenditure ranges from remarkably high to zero,
and the endowment of human resources seems to depend more on the denominator than
on the actual hiring of skilled personnel. Investments in innovation are biased, in most
cases towards the acquisition of capital goods, which is strongly related to access to
external sources of funding. As a results, their isolated investments are not enough to
increase their capability levels. This group of firms is also made up of enterprises from
different sectors, of different sizes, and different capital ownership. In this case, they are a
proof of the lack of direct association between a high-tech sector and a more high-profile
innovative strategy, as well as the possibility of finding large national firms (which are
supposedly in the best position to invest in innovation) that are basing their competitive
strategy on something other than innovation, most probably the exploitation of static
comparative (or even absolute) advantages.
Finally, there is the group of non-innovative firms. These are firms that did not invest in
knowledge for almost a decade, which had low interactions with the NIS and a reduced
level of skilled human capital. These firms survived the crisis and grew afterwards, proving
that innovation is one of many possible competitive strategies. These firms also prove that
macro incentives are not enough to trigger long-term dynamic innovative strategies.
To conclude, the innovative strategy of the firm is a source of micro-heterogeneity which
can complement the traditional size, sector and capital ownership classifications. It
accounts for firm-specific features regarding the innovation process but also for groups of
firms which lie outside the average values of standard classifications. At the same time, this
analysis has showed how firms can change the way they face the selection process. In the
following chapters, the impact of those different ways of competing will be analyzed in
order to study if the high-profile firms identified in this chapter are associated with better
economic performance.
- 203 -
Chapter 7: Innovative strategies, innovation
persistence, and firm performance
Introduction
This chapter aims to characterize the relationship between the firm’s strategies and its
performance. The hypotheses state that firms with a continuous innovative strategy
(continuous investments) tend to innovate persistently, which in turn leads to a better
economic performance. Since the model will be tested for the Argentinean case during
three different economic phases, it will allow for analysis of the way different strategies in
different macroeconomic environments impact both innovation results and firm growth. In
this way, the existence of firms better prepared to deal with the environment will be tested
for.
The results only partially confirm the hypotheses. Firms with a continuous strategy show
better performances, however, this is not because of a higher innovation persistence, but
because of higher probabilities of achieving positive results (present results depends
positively on continuity but negatively on past results). New innovative firms present worse
performances than continuous ones but higher than sporadic and non-innovative firms.
Sporadic firms show the poorest results in terms of the probability of achieving
innovations and a performance evolution closer to non-innovative firms than to the other
innovative strategies. As will be demonstrated, the classification based on the innovative
strategy of the firm allows for the analytical reduction of micro-heterogeneity and the
quantification and characterization of the impact of the different ways of seeking a
competitive advantage, which in turn accounts for different capabilities when dealing with
an unstable environment.
These findings will be presented as follows. After this introduction, in 7.1, the hypotheses,
the methodology, and the model are presented. Section 7.2 provides a descriptive analysis
of the relationship between the identified strategies and the firms’ performance. In 7.3, the
model is applied to the panel of Argentinean manufacturing firms and the results are
analyzed. Section 7.4 discusses the findings and the relationship with the literature. In 7.5,
some preliminary conclusions are provided.
Chapter 7: Innovative strategies, innovation persistence, and firm performance
- 204 -
7.1. Hypotheses and methodology
7.1.1. Hypotheses
According to what was presented in Chapter 3, the firm’s strategy is the set of decisions
regarding how to face the selection process, the impact of which is related to the firm’s
performance in the market. To learn whether some firms are better prepared to deal with
the environment, we need first to analyze how firms’ strategies impact their performance
and to what extent some strategies are “better” than others.
From Chapter 4, we concluded that the firm’s performance can be approximated by means
of the evolution of productivity, employment growth, and exports, on the assumption that
the impact of the innovative strategy on performance is mediated by the innovation results.
Productivity should account for the impact of the innovative strategy on the selection
process, in that specific innovations allow the firm to either reduce its cost or to sell
differentiated products with higher prices. Employment levels will be used as a measure of
growth, on the assumption that higher levels should account for an increase in production
levels. Exports are a common measure of competitiveness, on the assumption that access
to foreign markets is a sign of a competitive advantage, whether because of lower costs or
differentiated products.
Given the strategies, variables, and relationships presented in Chapter 5, and the
methodological questions presented in Chapter 4, two sets of hypotheses guide this
chapter. On the one hand, there are the hypotheses about innovation persistence,
associated to Q2:
Q2: Is there is a positive relationship between the continuity of the firm’s strategy and its
past and present innovation results? (Chapter 7)
H2.1: The probability of innovating depends positively on having innovated in the past.
H2.2: The probability of innovating depends positively on the innovative strategy.
H2.2.1: The probability of innovating depends positively on having a continuous
innovative strategy;
H2.2.2: The probability of innovating depends positively on having a new innovative
strategy;
H2.2.3: The probability of innovating depends positively on having a sporadic
innovative strategy.
Chapter 7: Innovative strategies, innovation persistence, and firm performance
- 205 -
H2.2.4: Continuous innovative firms have higher probabilities of innovating than the
rest of the strategies.
H2.3: The probability of innovating depends positively on having innovated in the
previous period when following a continuous innovative strategy.
H2.1 is what the literature about innovation persistence predicts (Chapter 3.3.2.1): past
innovations have a positive impact on present ones (a significant and positive coefficient in
the regressions for the lagged dependent variable). The group of hypotheses H2.2.1 to
H.2.2.3 establishes the expected positive relationship between present innovations and
strategy (a significant and positive coefficient for all innovative strategies). H2.2.4
establishes the expected ordering of the strategies, where the marginal effect of firms with a
continuous strategy is higher than the rest of the innovative strategies (new and sporadic
firms).
The theoretical rationale behind these hypotheses is attached to the findings of Chapter 6
and the dissertation made in Chapters 3 and 4: firms with a continuous innovative strategy
have a more complex innovative structure and, therefore, they are expected to have higher
probabilities of achieving innovations. Given that the findings from Chapter 6 provide
evidence of a change towards a more complex structure, new innovative firms are expected
to have higher probabilities of achieving innovations than firms with non-innovative and
sporadic strategies but lower than those with continuous ones. Finally, although firms with
a sporadic strategy show an innovative structure similar to that of non-innovative firms,
since these firms actually invested in innovation (at least in some sub-periods) should
increase their odds of achieving positive results.
H2.3 refers to the interaction between past results and the strategy (how past innovations
were achieved). In this case, the hypothesis aims to test the existence of an additional
probability of achieving results when the continuous innovative firm achieved innovations
in the past. The theoretical rationale behind this is that the more complex innovative
structure of continuous innovative firms should lead to more complex innovations, which
give the firms higher margins to seek complementary and marginal innovations. The
alternative hypothesis in this case is that, given the change in the environment, past
innovations could have had these characteristics but they do not fit the new environmental
situation, and the firm has to change its innovative trajectory in order to gain a competitive
advantage in the new environment. Therefore, the alternative hypothesis is that the lock-in
Chapter 7: Innovative strategies, innovation persistence, and firm performance
- 206 -
effect of past innovations decreases the odds of achieving results in the present (Chapter
3.3.3).
The second set of hypotheses refers to the relationship between the strategies and the
firm’s performance, which are associated with Q3:
Q3: Is there is a positive relationship between the continuity of the firm’s strategy, its
innovation results, and economic performance? (Chapter 7)
H3.1: Innovation results have a positive impact on firm performance (productivity, labor
growth and exports).
H3.1.1: Innovation results have a positive impact on the firm’s productivity levels;
H3.1.2: Innovation results have a positive impact on the firm’s employment levels;
H3.1.3: Innovation results have a positive impact on the firm’s export levels.
H.3.2: Continuous innovative firms improve their performance regardless of their
innovation results.
H3.2.1: Continuous innovative firms improve their productivity;
H3.2.2: Continuous innovative firms improve their employment growth;
H3.2.3: Continuous innovative firms improve their export growth.
H3.1.1 to H3.1.3 refer to the expected positive relationship between innovation results and
the firm’s performance, based on the three variables defined in Chapter 5 (productivity,
employment, and exports). This is what the literature predicts: the “innovation results”
variable is expected to have a positive and significant coefficient (Chapter 4.1.3).
H3.2.1 to H3.2.3 aim to test the positive impact of the strategy regardless of the innovation
results (the achieved innovations). The hypotheses refer to the expected positive impact of
the strategy (and the more complex structure) on the accumulation of capabilities and
resources arising from the innovation process, which in turn increase the firm’s
performance.
7.1.2. The model
7.1.2.1. General overview and the inclusion of strategies
The model I will use to test the hypotheses states that innovations of the firm at time
( depends on past innovations , the innovative strategy , and the
environmental determinants ) (Eq. 2—H3.1). As such, innovation will impact the
Chapter 7: Innovative strategies, innovation persistence, and firm performance
- 207 -
firm’s performance , which can be analyzed in terms of the evolution of the firm’s
labor productivity, total employment, and total exports (H3.2). The firm’s present
performance is a function of its past performance to account for its path
dependence, the innovations it has achieved , the innovative strategy ( , and
the environmental determinants ). (Eqs. 3.1 and 3.2)
(Eq. 2)
Where:
(Eq. 3.1)
(Eq. 3.2)
Between t and t-1, the cluster analysis performed in Chapter 6 confirmed the four types of
innovative strategies, based on the firm’s frequency of expenditure, which were also
correlated with different innovative structures (intensity and balance of expenditure,
qualified human resources, linkages, and access to external funds). These groups are:
where:
Continuous innovative firms (Cont): firms with sustained investments in
innovation;
Non-innovative firms (NI): firms with zero investments on innovation;
New innovative firms (New): firms which started to invest in innovation from one
sub-period to the following one;
Sporadic innovative firms (Spor): firms which stop investing in innovation from
one sub-period to the following one.
Consequently, accounts for four different groups of firms (continuous, new, sporadic,
and non-innovative firms), which vary over time. Common practice in the reviewed
literature on clusters and heterogeneity is to include strategies as dummy variables to see
how different behaviors change the average level of the estimations. In this case, the
strategies will be included as dummy variables and, in the case of innovation persistence,
also as an interactive (multiplicative) term. In theoretical terms, this solution implies that
the strategies determine how the innovation was sought and eventually achieved, and this
“how” is what makes the difference in terms of the innovation-related and economic
performance.
Chapter 7: Innovative strategies, innovation persistence, and firm performance
- 208 -
7.1.2.2. Innovation persistence
To test the H2 set of hypotheses, the model measures the impact of past innovations on
the probability of reaching innovations in the present, including the firm’s innovative
strategy. Since the dependent variable is a binary one (did or did not innovate at time t), a
dynamic random effects probit model was chosen, which also controls micro-heterogeneity
by means of the inclusion of unobserved effects. The selection of a random over a fixed
effects model responds to the inclusion of the Wooldridge’s (2005) solution to relax the
assumption regarding the independence between the observed and unobserved effects. At
the same time, there is no general transformation to eliminate unobservable effects in
binary estimations under fixed effects models. Finally, and from a theoretical point of view,
there are good reasons to believe that some unobservable effects are time-invariant and
firm-specific while others are time- and case-invariant. As a result, random effects better
suits the characteristics of the micro-heterogeneity to be controlled here: the innovation
persistence and the impact of innovations on performance.
Wooldridge’s (2005) solution consists of the inclusion of the explanatory variables in their
average value, and the initial condition (the value of the dependent variable at time zero).
Of course, except from the initial condition, this solution can only be applied to
continuous variables. However, the use of a random effect model allows for the inclusion
of dummy variables to control for the different time-invariant and case-invariant sources of
heterogeneity. This will be done by means of the environmental dummies.
Methodologically speaking, Wooldridge’s (2005) solution is based on Chamberlain’s (1980;
1984; 1992) model for random effects, adjusted to fit a dynamic estimation. The inclusion
of the initial condition and the independent variables at each point in time allows
assumptions about the lack of correlation between the idiosyncratic term, the error, and the
predictive variables to be abandoned. At the same time, it also allows Raymonds et al.’s
(2010) spurious persistence to be reduced, so the impact of unobserved or omitted
variables is minimized. Following the contributions of Peters (2009), Clausen et al. (2011),
and Raymond et al. (2010), and according to what Wooldridge explains, replacing these
variables with their average value minimizes the number of explanatory variables without
affecting the results of the estimation. 92
92 Panos (2008) tested Heckman, Wooldridge, and Orme’s solutions for dynamic probit models and found that each of them provide similar results, with no evidence of one estimator being superior to another.
Chapter 7: Innovative strategies, innovation persistence, and firm performance
- 209 -
Theoretically speaking, Wooldridge’s solution is based on the assumption that a firm’s
unobserved characteristics can be approximated with a linear function of its observable
behavior. This implies that the firm’s innovative structure (linkages, expenditures, qualified
human resources, and access to external funds, taken into account together) is in part the
result of its unobservable characteristics. In terms of performance, it means that the
historic productivity levels, employment, and exports are also the result of the firm’s
idiosyncratic features and therefore, they can also control these features.
The endogeneity problem that could arise from the interaction between innovations results
and innovation activities is one limitation that this solution cannot overcome. Dynamic
probit models are run on the assumption that the explanatory variables are strictly
exogenous. In the present case, this would imply that innovations do not impact the firm’s
capabilities, future investments, or financial situation. In this respect, Coad and Rao (2011)
found that within short periods, this is a valid assumption given the time that it takes for
innovation investments to impact on performance and so to impact on other dimensions
of the firm’s innovative structure. However, since dynamic models control serial
correlation, even when this type of endogeneity actually exists, coefficients have been
proved to be consistent (Wooldridge, 2005).
Finally, in order to analyze the theoretical contributions of the innovative strategies and
establish comparisons with the literature on persistence, this phenomenon will be tested
with and without differentiating between innovative strategies.
The model to test the set of hypotheses H2 can be formalized as follows:
(Eq. 2)
(Eq. 2.1)
(Eq. 2.2)
=
(Eq. 2.3)
(Eq. 2.4)
Where:
According to Eq. 2.1, the innovation of the firm at time ( depends on
innovations in t-1 the innovative strategy , the interaction between past
Chapter 7: Innovative strategies, innovation persistence, and firm performance
- 210 -
innovations and the strategy a set of observable attributes case- and
time- variant, referred to the firm’s innovative structure ; the
environmental determinants, some of them time-invariant, some of them case-invariant,
but also observable ; and a set of idiosyncratic characteristics,
unobservable and time-invariant . The unobservable characteristics are the
initial condition plus the rest of the continuous explanatory variables (Wooldridge
solution). Non-innovative firms will be set as the reference group and correspond to
the statistical errors. Table 7.1 summarizes each variable.
Table 7.1: Innovation persistence: summary of the dimensions and variables
Label Detail Calculation Value
Observable characteristics
Product, process, organization, or
commercialization innovations At least one innovation in the sub-period. 0;1
Innovative strategies Continuous, new, sporadic, and non-innovative
firms. One dummy per strategy. 0;1
Innovation intensity Innovation expenditure to sales. Annual average for
the sub-period. 0–
Innovation balance Distribution of the total innovation expenditure. 0–
Qualified human resources Number of professionals to employment. Annual
average for the sub-period. 0–
Linkages At least one linkage in the sub-period. Variable
available for t and t-2. 0;1
Access to external sources of funding At least one external source in the sub-period. 0;1
Total employment Total employment, average value for the sub-period. 0–
Capital ownership More than 1% of their shares owned by foreign
capital in t-1 0;1
Sectoral belonging (2-digit ISIC, in t-1) One dummy for each sector. 0;1
Sub-period One dummy for 0;1
Unobservable characteristics
Initial condition in t-2 0;1
Average of observable characteristics Average values of the corresponding variables 0–
7.1.2.3. Strategies, innovation, and performance
To test the differential impacts of innovations on the firm’s performance, subject to the
innovative strategy (H3), productivity, employment, and export intensity will be analyzed
by means of a dynamic OLS regression. To control for the potential endogeneity between
innovations and performance, the CDM solution was adopted (Crépon, et al., 1998).
According to Crépon et al. (1998), the potential endogeneity between innovation results
and firm performance (a bidirectional relationship) can be solved by using instrumented
variables. The authors propose instrumenting innovation through innovation expenditure
Chapter 7: Innovative strategies, innovation persistence, and firm performance
- 211 -
and a set of the firm’s innovation characteristics, plus the traditional structural controls
(sector, size, and capital ownership). To instrument the innovation results, the predicted
values for the variable from Eq. 2.1 without the inclusion of the strategies were
used.
This adaptation of the CDM solution states that innovations at time t impact performance.
Since the definition of innovation includes its successful introduction into the market, it
must impact sales (or costs) within the same time. Consequently, innovations must be
included in the same sub-period, together with performance.
The model can be formally written as follows and Table 7.2. summarizes each variable:
(Eq. 3)
(Eq. 3.1)
(Eq. 3.2)
(Eq. 3.3)
Where
)
The model states that the firm’s performance ) can be analyzed in terms of
productivity, growth, and exports ( respectively) and it is a function of its
past performance , the innovative strategy the innovations it has achieved
, the environmental determinants ), and the unobserved effects ( .
Following the CDM solution, the endogenous variable (innovations) will be instrumented
as the predicted probability of innovating ( ) obtained in equation 2.1, used to test
persistence. Unlike in the original CDM model, past performance will be also included, to
control for path dependence. Unobserved effects ( will be controlled with the
Wooldridge (2005) solution in the same way as was done for the persistence analysis. Firm
performance will be approximated using productivity , employment growth93 ( ,
93 To test the relationship between the strategies and firm growth, average sales for each sub-period will be used as a control variable for size, thus replacing the use of employment for this function.
Chapter 7: Innovative strategies, innovation persistence, and firm performance
- 212 -
and exports . In the case of exports, the average export level for the sub-period
( will be set as the dependent variable and a dynamic random effect regression will
be run, as with productivity and employment, but controlling for the condition of whether
the firm is an exporter or not ( if the firm exported at time ).
Table 7.2: Performance: summary of the dimensions and variables
Label Detail Calculation Value
Observable characteristics
Predicted innovation results ( Eq. 2.1—Chapter 6 0;1
Innovative strategies. Continuous, new, sporadic, and non-innovative
firms. One dummy per strategy. 0;1
Performance
Productivity Total sales to employment, annual average for the
sub-period. In natural log. 0–
Growth Total employment of the firm, annual average for
the sub-period. In natural log. 0–
Exporter firm Exports > 0 in at least one year of the sub-period. 1;0
Exports Total exports, annual average for the sub-period.
In natural log. 0–
Total sales Total sales, average value for the sub-period. 0–
Capital ownership More than 1% of shares owned by foreign capital
in t-1 0;1
Sectoral belonging (2-digit ISIC, in t-1) One dummy for each sector. 0;1
Sub-period One dummy for 0;1
Unobservable characteristics (
Initial condition Value in t-2 of the corresponding variable 0–
Average of observable characteristics Average values of the corresponding variables 0–
7.2. Results from the descriptive analysis
Table 7.3 summarizes the probability of achieving innovations for each combination of
strategies.94 As expected, besides the higher levels observed among the continuous firms
and the lower levels among the non-innovative, innovation results seem more affected by
the strategy in each sub-period than by the combination of strategies. A high correlation
between the initial situation (innovations in 1998–2001) and the last sub-period (2005–
2006) is also observed in all cases: there is positive association of innovations in t and t-2
among the continuous and new innovative, and a negative one among the non-innovative
and the sporadic groups.
94 Statistical tests over the mean are based on the same methodology and tests used in Chapter 6.1.2.1.
Chapter 7: Innovative strategies, innovation persistence, and firm performance
- 213 -
Table 7.3: Innovation persistence and strategies——% of firms (std. residuals)
Sub-period
Inno
1998–2001 2002–2004 2005–2006
Innovative Continuous Continuous
94.15* (8.3) 48.31* (2.4) 81.54* (7.3)
Sporadic
80.49 (1.8) 51.22 (1.2) 2.44* (-4.4)
Sporadic New
84.85* (2.0) 42.42 (0.2) 63.64 (0.9)
NI
85.11* (2.4) 34.04 (-0.6) 4.26* (-4.5)
Non-Innovative New Continuous
31.67* (-3.9) 55.00* (2.6) 76.67* (3.7)
Sporadic
4.00* (-3.6) 44.00 (0.3) 4.00* (-3.3)
NI New
14.93* (-4.7) 47.76 (1.0) 52.24 (0)
NI
9.86* (-7.6) 0.70* (-7.4) 0* (-8.6) Obs.: 800. Inno: at least one innovation during the sub-period, percentage of firms. *: significant at 99%, sign given by the sign of the std. residuals (Pearson’s Chi-square and Kendall’s Tau-c tests). See Appendix 3.1.1. Source: own elaboration based on INDEC (2010).
The evolution in productivity levels is presented in Table 7.4. In it and the tables that
follow, the lack of basic controls distorts the results, which is observed in the standard
deviation; however, mean values provide an initial approach to the data. The differences
between strategies remain high and the C-C firms show the highest levels of sales per
employee, followed by the N-C group, which stands out from the rest of the firms from
the beginning of the period. Although the lack of size controls could be part of the
explanation of these differences, the higher levels among the continuous and new
innovative firms and the significantly lower levels among the non-innovative ones
constitute a first approximation to the differential performance of the more complex
innovative strategies.
Chapter 7: Innovative strategies, innovation persistence, and firm performance
- 214 -
Table 7.4: Productivity and innovative strategies (ARS 000)—Mean (std. dv.)
Sub-period
Pd (ARS 000)
1998–2001 2002–2004 2005–2006
Innovative Continuous Continuous
118 (153) 142 (222) 153 (243)
Sporadic
105 (178) 113 (237) 117 (229)
Sporadic New
78 (73) 74 (97) 90 (97)
NI
100 (105) 96 (125) 107 (183)
Non-innovative New Continuous
100 (118) 107 (124) 114 (139)
Sporadic
65 (49) 68 (63) 72 (67)
NI New
79 (86) 91 (105) 103 (107)
NI
54 (65) 53 (76) 60 (77)
Obs.: 800. Pd: sales per employee, average levels for the sub-periods in ARS 000. KW and JT positive and significant at 99% for the eight combined strategies. Complete estimations in Appendix 3.1.2. Source: own elaboration based on INDEC (2010).
The evolution of the different innovative strategies in terms of employment is presented in
Table 7.5. The first observation is that large firms bias the results, which is why tables are
presented for the total panel and for SMEs (less than ARS 200 million).
Of course, the most obvious finding is the high impact of large firms, which elevates the
average number of employees per firm among the C-C. When large firms are removed, C-
C and N-C firms present values closer to the rest of the panel, although they remain those
with the highest levels, which are also sustained and improved over time. Non-innovative
and sporadic innovative firms present similar levels of employment, also sustained over
time, except for the NI-NI, where a slight drop is observed over the three sub-periods.
Chapter 7: Innovative strategies, innovation persistence, and firm performance
- 215 -
Table 7.5: Employment—number of people—Mean (std. dv.)
Sub-period Sub-period
Gr (Total Panel)
1998–2001 2002–2004 2005–2006
Gr (SMEs)
1998–2001 2002–2004 2005–2006
Innovative Continuous Continuous Innovative Continuous Continuous
309 (634) 291 (561) 319 (546) 192 (274) 181 (227) 207 (246)
Sporadic Sporadic
111 (108) 104 (97) 112 (108) 98 (84) 93 (81) 97 (85)
Sporadic New Sporadic New
96 (118) 79 (87) 88 (83) 91 (116) 72 (78) 83 (79)
NI NI
108 (120) 88 (105) 93 (90) 106 (120) 85 (105) 91 (90)
Non-innovative
New Continuous
Non-innovative
New Continuous
144 (197) 140 (197) 170 (235) 111 (102) 100 (90) 127 (118)
Sporadic Sporadic
53 (49) 52 (54) 65 (79) 53 (49) 52 (54) 65 (79)
NI New NI New
79 (104) 77 (117) 82 (109) 79 (104) 77 (117) 82 (109)
NI NI
58 (63) 51 (65) 54 (68) 58 (63) 51 (65) 54 (68) Total panel: obs. 800. SMEs (small: obs. 756. Gr: total employment, average values for the sub-periods. KW and JT positive and significant at 99% for the eight combined strategies, with and without sectoral controls. Ordering: continuous, new, sporadic, and non-innovative. The ordering provided by average values is the same as the one provided by the rankings. Complete estimations in Appendix 3.1.3 and 3.1.4. Source: own elaboration based on INDEC (2010).
Finally, Table 7.6 presents the average probability of the firm being an exporter (Xr) and
the average level of exports (total exports per firm per year) for each group of firms. In this
case, differences between strategies are clearer, with higher probabilities of continuous and
the new innovative firms becoming exporters (or continuing to export), although with a
decreasing tendency.
The total level of exports (Xpo) increases over time regardless of the firm’s strategy,
although the highest increase is observed among the more dynamic innovative strategies
(C-C and N-C). This evolution is consistent with the general trends identified for 2002–
2006, which was defined as an export-led model of growth (see Chapter 2). As we shall see
in the econometric estimations, even when size and sector are controlled, firms with a
more dynamic strategy show a significantly better performance in terms of exports, which
coincides with the distance in the levels between continuous and new firms, on the one
hand, and sporadic and non-innovative enterprises, on the other.
Chapter 7: Innovative strategies, innovation persistence, and firm performance
- 216 -
Table 7.6: Exports—% of firms and intensity
Sub-period—% firms (std. residuals)
Sub-period—Mean (std. dv.)
Xr (% of firms)
1998–2001 2002–2004 2005–2006
Xpo (ARSMM)
1998–2001 2002–2004 2005–2006
Innovative Continuous Continuous
Innovative Continuous Continuous
83.08* (3.5) 80.62* (3.3) 74.46* (4.3)
7.05 (26.87) 11.44 (46.55) 28.18 (132.49)
Sporadic
Sporadic
68.29 (0.1) 65.85 (0) 56.10 (0)
8.34 (31.77) 10.85 (46.42) 24.41(109.24)
Sporadic New
Sporadic New
60.61 (-0.5) 66.67 (0) 48.48 (-0.6)
0.52 (0.97) 0.59 (1.35) 2.32 (6.71)
NI
NI
61.70 (-0.5) 51.06 (-1.3) 44.68 (-1.1)
1.67 (4.01) 1.46 (4.35) 3.17 (10.36)
Non-Innovative
New Continuous
Non-innovative
New Continuous
73.33 (0.8) 72.50 (1.0) 66.67 (1.5)
4.37 (14.33) 7.57 (25.89) 17.47 (72.6)
Sporadic
Sporadic
52.00 (-1.0) 44.00 (-1.4) 36.00 (-1.4)
0.51 (1.40) 0.69 (2.18) 1.00 (2.96)
NI New
NI New
49.25 (-1.8) 56.72 (-0.8) 41.79 (-1.6)
0.56 (1.87) 0.63 (2.1) 2.07 (4.75)
NI
NI
41.55* (-3.8) 38.03* (-4.1) 23.94* (-5.2)
0.47 (2.07) 0.53 (2.93) 1.14 (5.71) Obs. 800. Xr=1 if exports>0 during the sub-period, % of firms. *: significant at 99%, sign given by the sign of the std. residuals (Pearson’s Chi-square and Kendall’s Tau-c tests). Xpo: average annual exports for each sub-period, ARS millions, constant prices 1998. KW and JT positive and significant at 99% for the three sub-periods and the eight combined strategies. See Appendix 3.1.5. and 3.1.6. Source: own elaboration based on INDEC (2010).
7.3. Estimation of the model
7.3.1. Innovation persistence
Table 7.7 presents the estimation of the model to test the relationship between persistence
of innovation and innovative strategies. In all cases (Models I.1 to I.3), the results show
that persistence is not corroborated (past innovations’ coefficients are not significant),
while the innovative structure shows a positive and significant impact on innovations for
the case of innovation balance, access to external resources, and linkages with the NIS.
When the innovative strategies are included (Model I.2), persistence remains non-
significant, while the strategies show a positive and significant impact on present
innovations. The interaction between the strategies and past innovations (Model I.3) shows
insignificant correlations except for the case of continuous innovative firms and past
innovations. In this case, the interaction is significantly negative, which means that past
innovations decrease the odds of achieving innovations in the present. Among sporadic
innovative firms, the loss of significance of the coefficient when the strategy is considered
both in isolation and interacting with past innovations means that this strategy does not
Chapter 7: Innovative strategies, innovation persistence, and firm performance
- 217 -
increase the odds of achieving innovations in the present (these firms have the same odds
as non-innovative firms).
Although the marginal effects must be interpreted with caution (see Chapter 5.3), results
allow the following observations to be made. Based on Model I.3 (Table 7.8), continuous
innovative firms have, on average, a 48% higher probability than non-innovative firms of
achieving innovations. Among new innovative firms, this percentage decreases up to 34%,
and the coefficient for sporadic firms is insignificant, which means that these firms face the
same probabilities as non-innovative firms. As most of the determinants of innovation are
being controlled, it is the isolated impact of the strategy that is being looked at, which
implies that—regardless of expenditure levels, capabilities, and funding—firms with the
highest frequency have almost 15% more probability of succeeding than firms that invested
only at the end of each sub-period (continuous versus new innovative firms).
Regarding the negative sign of the interaction between the continuous strategy and the past
results (Model I.3), the final marginal effect of the strategy is positive and significant95,
which means that despite the decrease in their probabilities attached to past innovations,
the impact of a sustained innovative strategy remains positive. If one includes the fact that
the period covers three distinct macroeconomic moments, these results mean that in the
case of continuous firms there is a lock-in effect associated with past innovations but the
accumulated capabilities and resources have allowed them to successfully react to the
changes and adjust the innovative strategy in a way that the probabilities of success remain
high.
Regarding the innovative structure96, the results differ notably between the three models.
When strategies are not included, the balance of innovation expenditure, linkages with the
NIS, and access to external resources have a significantly positive sign: moving towards a
more balanced investment in innovation (combining sources of knowledge) increases the
odds of achieving innovations by almost 33% (average values for the average case in
comparison with non-innovative firms). Establishing linkages with the NIS has a similar
impact (31%). Although this dummy alone does not allow conclusions to be drawn
regarding the interaction between the firm and the NIS, it does provide support for the
95 Since the multiplicative term is made up of two variables included in the model, the final marginal effect arises from the combination of odds for each variable (positive and negative in this case), which is why this term has not been included in Table 7.8. 96 It is important to keep in mind that the variables included (expenditures, capabilities, and resources) are controlling the innovative structure. They do not tend to measure the impact of innovation investments on present results.
Chapter 7: Innovative strategies, innovation persistence, and firm performance
- 218 -
hypothesis in terms of the positive impact of accessing external sources of knowledge,
regardless of the type of organization and the degree of formalization of the link (formal
cooperation versus interactions—see Chapter 1.2.2.4).
Table 7.7: Random effects dynamic probit model—Dep. variable: Innovations (Inno)
Model I.1 Model I.2 Model I.3
Coeff. S.E. Coeff. S.E. Coeff. S.E.
Structural equation
-0.03530 0.19352 -0.43196 0.30652 -0.04742 0.38375
1.47610*** 0.30681 2.29776*** 0.44915
1.26771*** 0.19364 1.27017*** 0.23710
0.70896*** 0.24763 0.51590 0.38781
* -1.2211*** 0.43890
* -0.04023 0.40881
*
0.01185 0.50296
1.31278 1.03342 0.36659 1.03742 0.68116 1.21462
1.09379*** 0.28848 0.34510 0.34193 0.42468 0.38399
-0.48655 0.39077 -0.35284 0.42410 -0.50997 0.48783
1.061838*** 0.13938 1.041811*** 0.15503 1.150139* 0.17081
0.3585498** 0.13167 0.2933045** 0.14483 0.2916018* 0.16489
* * *
0.00019 0.00060 0.00029 0.00066 0.00039 0.00073
-0.16389 0.11570 -0.19446 0.13488 -0.22554 0.16225
0.12374 0.10182 0.2520474** 0.13060 0.488980*** 0.16750
-1.03849*** 0.20779 -1.72349*** 0.31631 -2.0988*** 0.40163
Individual heterogeneity
0.06728 0.15052 0.12909 0.17895 0.12337 0.19312
2.993251** 1.49309 1.41562 1.65153 1.75048 1.92774
0.9348673** 0.40845 0.51905 0.47943 0.57810 0.54840
0.81839 0.74053 0.84930 0.83857 1.09639 0.98565
0.00012 0.00059 0.00003 0.00064 0.00002 0.00071
-2.42679 1.88607 -1.16827 0.96557 -0.28730 0.61081
0.29719 0.28026 0.55759 0.26920 0.86619 0.26454
0.08115 0.14064 0.23717 0.17469 0.42866*** 0.14959
Obs.: 2400. ***, ** and *: significance at 99%, 95%, and 90%, respectively. =1 if at least one innovation
was achieved during the sub-period. time-average value of the corresponding variable.
omitted due to collinearity, = 2002–2004; = linkages 2005–2006. = sectoral controls included. Estimations based on Gauss–Hermite, using twelve quadrature points. Quadrature checks, no variations over 1%. Complete estimations in Appendix 3.2. Source: own elaboration based on INDEC (2010).
Chapter 7: Innovative strategies, innovation persistence, and firm performance
- 219 -
Table 7.8: Marginal effects - RE dynamic probit—Dep. variable: Innovations (Inno)
Model 1 Model 2 Model 3
Marg. Eff. S.E. Marg. Eff. S.E. Marg. Eff. S.E.
Structural equation
-0.01038 0.05657 -0.11397 0.07386 -0.16834 0.05814
0.432360*** 0.06645 0.481924*** 0.07345
0.361433*** 0.04351 0.347731*** 0.05716
0.177846*** 0.05480 0.12353 0.06065
0.38605 0.30342 0.09672 0.27354 0.16321 0.29008
0.321655*** 0.08146 0.09105 0.08887 0.10176 0.09006
-0.14308 0.11407 -0.09310 0.11108 -0.12219 0.11518
0.312259*** 0.03277 0.274879*** 0.03173 0.2755831* 0.03279
0.10544** 0.03794 0.077388** 0.03722 0.0698703* 0.03867
* * *
0.00006 0.00018 0.00008 0.00017 0.00009 0.00018
-0.04820 0.03385 -0.05131 0.03507 -0.05404 0.03825
0.03639 0.02932 0.066502** 0.03175 0.1171638* 0.03407
Individual heterogeneity
0.01979 0.04376 0.03406 0.04570 0.02956 0.04504
0.8802361** 0.43134 0.37351 0.43432 0.41943 0.45934
0.2749198** 0.11748 0.13695 0.12779 0.13852 0.13272
0.24067 0.21626 0.22409 0.21838 0.26270 0.23157
0.00004 0.00017 0.00001 0.00017 0.00000 0.00017
Obs.: 2400. ***, ** and *: significance at 99%, 95%, and 90%, respectively. =1 if at least one innovation
was achieved during the sub-period. time-average value of the corresponding variable. = 2002–
2004; = linkages 2005–2006. = sectoral controls included. Margins: delta method at
. Complete estimations in Appendix 3.2.
Source: own elaboration based on INDEC (2010).
Another interesting result is the positive and significant impact of the average expenditure
on innovation ( ) in Model I.1. Although this variable is just a proxy of the firm’s
unobserved characteristics, its positive and significant coefficient (a 1% increase in the
expenditure increases the probabilities of positive results by 88%) indicates that the
duration of innovation projects does not necessarily correspond to the sub-periods
identified here. Similarly, it confirms the importance of sustained investments in these
activities (a continuous strategy).
Surprisingly, the level of qualified human resources does not increase the odds of achieving
innovations in any of the models. Although these results coincide with those of a similar
study (Huergo and Moreno, 2011), it contradicts the finding of several authors based on
numerous empirical studies, both for developed and developing countries (Kemp et al.,
2003; Chudnovsky et al., 2004; Borello et al., 2006; Milesi, 2006; Jensen et al., 2007;
Lugones et al., 2007; Lugones et al., 2008; Dutrenit and Puchet, 2011). From Chapter 6 we
know that the variance in qualified human resources accounts for a low variance among the
most dynamic innovative strategies, which are, in turn, the ones with the highest levels of
Chapter 7: Innovative strategies, innovation persistence, and firm performance
- 220 -
skilled personnel. Therefore, one possible explanation for this lack of significance could be
that the impact of this variable is absorbed by the rest of the innovative structure
(especially IB) among the more dynamic firms and that it is not enough to trigger
innovations among firms with a low-profile innovative behavior (the different sporadic
innovative firms). That is, it is not significant in terms of present results. In any case,
further studies should shed light on this lack of correlation.
Access to external resources (ER) impacts positively on the probability of achieving
innovations. In this case, firms that managed to access external sources of funding have a
10% higher probability of achieving innovations in the present (Model I.1). However, even
though this impact is significant, it is lower than that of the rest of the variables and the
coefficient decreases as the strategies are included, which could be associated with the fact
that Argentinean firms rely on their own funds to source innovation (Chapter 2.3).
In Models I.2 and I.3, only Link and RE remain significant, most probably due to the
impact of the strategies, which, as analyzed in Chapter 6, are highly correlated with specific
innovative structures.
7.3.2. Productivity and innovative strategies
The estimation of the relationship between productivity and strategies is presented in Table
7.9 (Models P.1 and P.2). The main result is the lack of correlation between productivity
and innovation results ( ), which was instrumented based on Model I.1. Regardless of
the inclusion of the strategies, innovations do not impact the level of productivity, at least
not when they are observed simultaneously.97 Although this result is not consistent with the
general literature on innovation, it is in keeping with the literature on Latin American
countries, for which several authors sustain that despite the high rate of firms with positive
innovation results (innovations), the scope of innovations is reduced, and thus so is the
impact they have on productivity (Dutrenit and Katz, 2005; Anlló et al., 2007; Lugones and
Suárez, 2010).98 This also begs the question of what respondents interpreted as innovation
and to what extent this variable can be compared across countries.
97 One of the assumptions of this estimation is the simultaneous (without lag) impact of innovations on productivity, which is also the generalized assumption of the literature, based on the idea that the definition of innovation includes the successful introduction of the innovation into the market. However, even if the variable innovations is lagged one sub-period, the coefficient remains non-significant. 98 One textbook example of this situation is the introduction of a new machine close to the technological frontier, without further investments on its selection and adaptation to the firm´s conditions. When the firm is below the frontier (such is the case of many Argentinean manufacturing firms), the low articulation with
Chapter 7: Innovative strategies, innovation persistence, and firm performance
- 221 -
Model P.2 aims to quantify the impact of the strategies. In this case, the impact of the
more complex strategies is positive and significant: the continuous and the new innovative
firms have positive and significant coefficients that are similar to one other. Given
productivity in t-1, productivity in t increases by 0.125 and 0.131 among continuous and
new innovative firms, respectively, (ceteris paribus the rest of the variables and if the rate is
constant). Among sporadic innovative firms, the relationship is insignificant, which means
that this strategy has the same impact on productivity as the non-innovative strategy.
These findings support the idea that the impact of the innovation process on the firm’s
performance depends on how the innovation process was carried out. The higher
coefficients among the continuous and the new innovative firms can be explained by the
accumulation of knowledge and capabilities, which is the result of more dynamic
innovative behavior (the higher levels of innovation investments, qualified human
resources, and linkages, observed in Chapter 6). As a consequence, the accumulation of
knowledge and enhanced capabilities allowed the firm to improve its routines, which in
turn improved productivity levels. In this case, the results are in line with the literature and
they also coincide with what was observed in the descriptive statistics, but in terms of the
positive impact of the innovation process on the firm’s performance. This reinforces the
idea of the impact of a sustained strategy on the firm’s performance, regardless of the
results, in that the way innovations were sought is what triggers accumulation processes
that enhances firms’ capabilities and, in terms of Penrose (1959), the performance of the
resources.
Finally, another two variables that significantly impact productivity levels are firm size
(although its impact is small) and capital ownership.99 With regard to the former, since the
previous level of productivity is controlled, the negative association is explained by the fact
that smaller firms tend to gain productivity faster than larger ones and also that
productivity improvement tends to have a large marginal impact when productivity is low.
In terms of capital ownership, the results show that foreign-owned firms reach higher
productivity levels. In this case, one possible explanation is the fact that these firms operate
in the local market but look to the regional one (Porta et al., 2011). The 2002 devaluation
increased the competitiveness of local production, which could have led to an increase of
local subsidiaries in the regional market share. Conversely, a second possible explanation is
the firm´s productive function reduces the impact of the new acquisition, even to the point that it does not improve the productive process at all. 99 See Appendix 3.3.1 for the analysis of different inclusions of the variables.
Chapter 7: Innovative strategies, innovation persistence, and firm performance
- 222 -
that this is due to the expansion of domestic demand together with an increase in the cost
of imports, which led these firms to increase their local production so as to respond to the
change in the domestic market. In any case, since data on the strategies of firms owned by
foreign capital is, by definition, not included in national innovation surveys, only future
research could shed light on the characteristics of such strategies and the direction of the
change (if any) in these firms’ innovative behavior.
Table 7.9: Random effects dynamic regression model—Dep. var.: Productivity (Pd)
Model P.1 Model P.2
Coeff. S.E. Coeff. S.E.
Structural equation
0.6794892*** 0.05689 0.663108*** 0.05629
0.02205 0.01444 -0.01280 0.01610
0.125211*** 0.03120
0.131515*** 0.03422
0.01797 0.03828
* *
0.000000001** 0.00000 0.000000001** 0.00000
0.0548971** 0.02646 0.0549075** 0.02645
-0.1051959*** 0.02125 -0.1151297*** 0.02114
0.2825132*** 0.07289 0.2371292*** 0.07397
Individual heterogeneity
0.2465215*** 0.06261 0.2539476*** 0.06188
-0.000000001* 0.00000 -0.000000001* 0.00000
0.00000 0.00000
0.29769 0.29744
0.00000 0.00000
Within 0.0033 0.0017
Between 0.9133 0.9142
Overall 0.8328 0.8353
Obs.: 2400. ***, **, and *: significant at 99%, 95%, and 90%, respectively. : sales to employment, average
value for the sub-period, in ARS 000, constant prices 1998, natural log. time-average value of the
corresponding variable. omitted due to collinearity, = 2002-2004; = sectoral controls included. Robust standard errors. Complete estimations in Appendix 3.3.1. Source: own elaboration based on INDEC (2010).
7.3.3. Employment growth and innovative strategies
Unlike the results of the previous section, innovations have a significant and positive
impact on the firm’s employment levels (Table 7.10). Models G.1 and G.2 confirm that the
innovations achieved lead to higher employment levels. The preliminary conclusion that
can be drawn from these results is the lack of confirmation of the hypothesis about
technological unemployment (Coad and Rao, 2011; Chapter 4.2.2). On the contrary,
innovations lead to increases in employment levels, regardless not only of the type of
Chapter 7: Innovative strategies, innovation persistence, and firm performance
- 223 -
innovation but also of the type of strategy: on average and given the employment level in t-
1, firms that achieved innovations reached an increase of 0.75 in the employment level in t.
These results seem consistent with the expansion of domestic demand during the period
under analysis, after the 1998–2001 crisis. At the same time, if one analyzes the results
taking into account the non-significant impact of innovations on productivity, it seems that
the innovations achieved during the period aimed to expand production capacity rather
than increasing productivity levels.
In respect of the strategies, continuous innovative firms managed to grow faster than the
rest regardless of the innovation results, and that explains the higher levels of total
employment observed in the descriptive statistics. On average and given the employment
level in t-1, firms with this strategy show an increase of 0.14 in the employment level in t.
For the rest of the strategies, the coefficients are non-significant, meaning that the impact
of these strategies is similar to that of non-innovative strategies. Therefore, firms with a
sporadic strategy do not have higher probabilities of achieving results, or of increasing their
productivity or production levels (employment) than non-innovative strategies. Firms with
a new innovative strategy improve productivity but this does not impact growth, at least
not yet. As the impact of the achieved innovations is positive and significant, the lack of
significance of the coefficient of these strategies means that in both groups of firms (new
and sporadic), the impact of the innovation process on the employment level depends on
the result of that process, in other words, it depends on being successful in innovating.
If these results are analyzed in relation to the higher participation of qualified human
resources among continuous innovative firms, together with the higher productivity levels,
these firms can be expected to pay, on average, higher salaries. Conversely, given the low
intensity of qualified human resources and the lower impact on productivity, the
employment evolution among the sporadic group of firms raises questions regarding the
quality and stability of the jobs created.
Chapter 7: Innovative strategies, innovation persistence, and firm performance
- 224 -
Table 7.10: Random effects dynamic regression model—Dep. variable: Growth (Gr)
Model G.1 Model G.2
Coeff. S.E. Coeff. S.E.
Structural equation
0.75556*** 0.04798 0.74459*** 0.04865
0.12271*** 0.01739 0.083617*** 0.01779
0.14108*** 0.03233
0.06075 0.03843
-0.02334 0.04848
* *
0.00000 0.00000 0.00000 0.00000
0.04624 0.03140 0.04125 0.03185
-0.14771*** 0.02076 -0.15255*** 0.02076
0.46483*** 0.08858 0.439682*** 0.09033
Individual heterogeneity
0.13703*** 0.04569 0.1364945* 0.04595
0.00000 0.00000 0.00000 0.00000
0.00000 0.00000
0.25384 0.25268
0.00000 0.00000
Within 0.00590 0.00530
Between 0.93640 0.93670
Overall 0.87580 0.87760
Obs.: 2400. ***, ** and *: significant at 99%, 95%, and 90%, respectively. : employment, average value for
the sub-period, natural log. time-average value of the corresponding variable. omitted due to
collinearity, = 2002-2004; = sectoral controls included. Robust standard errors. Complete estimations in Appendix 3.3.2. Source: own elaboration based on INDEC (2010).
7.3.4. Exports and innovative strategies
The impact of innovations on exports is positive and significant with and without the
inclusion of the innovative strategies (Table 7.10, Models X.1 and X.2). Regarding the
strategies, the econometric analysis confirms what was observed in the descriptive statistics:
firms with a continuous or a new innovative strategy reach higher export levels. Given the
export level in t-1, continuous and new innovative firms reach, on average, an export level
0.33 and 0.42 higher in t, respectively, and ceteris paribus. As with the analysis performed on
the other two performance variables, the characteristics of these innovative strategies
trigger additional effects, which—in keeping with what was discussed in Chapters 3 and
4—are associated with the increase in the firm’s capabilities to exploit its resources. The
higher impact of the new innovative strategy could be explained by the fact that these firms
began an innovative strategy once the environment changed (2002–2004) so they were
better prepared to take advantage of the new relative prices (a competitive exchange rate)
and managed to exploit the competitive shock of the devaluation.
Chapter 7: Innovative strategies, innovation persistence, and firm performance
- 225 -
With regard to the rest of the variables, in both estimations, the capital ownership and the
size of the firm also have a positive and significant impact on export levels, but the
coefficient of the variable size is remarkably low (which might be due to the correlation
between firm size and capital ownership).
Now that the difference performance variables have been examined, it can be argued that
firms with a continuous innovative strategy reach higher productivity, employment, and
export levels, which support the hypothesis of a virtuous growth process. This process
seems to be able to contribute to relaxing external restrictions by means of higher export
levels; to overcome the persistence of an almost 10% unemployment rate by increasing the
employment levels; and to improve the income levels through higher productivity levels.
Of course, only future research could confirm or reject and characterize this hypothesis
and improve the characterization of the “best” strategies to deal with the Argentinean
challenges discussed in Chapter 2.
Table 7.11: Random effects dynamic probit model—Dep. variable: Exports (Xpo)
Model X.1 Model X.2
Coeff. S.E. Coeff. S.E.
Structural equation
0.08315*** 0.01419 0.07894*** 0.01415
0.26689*** 0.05950 0.21203*** 0.06192
0.33050** 0.14175
0.42099*** 0.12833
0.18091 0.13713
* *
11.25807*** 0.20671 11.26248* 0.20468
0.0000001* 0.00000 0.0000001** 0.00000
1.210392*** 0.18226 1.210301*** 0.18172
-0.65341*** 0.05934 -0.67072*** 0.05943
-0.3222 0.21540 -0.51236** 0.22870
Individual heterogeneity
0.11600*** 0.01774 0.1156176*** 0.017723
0.0000001*** 0.00000 0.0000001*** 0.00000
1.29749 1.30199
1.07299 1.06526
0.59386 0.59901
Within 0.8893 0.8915
Between 0.9448 0.9446
Overall 0.9389 0.9389
Obs.: 2400. ***, ** and *: significant at 99%, 95%, and 90%, respectively. =total exports, annual average
for the sub-period, ARS 000 in constant prices 1998, natural log. time-average value of the corresponding
variable. omitted due to collinearity, = 2002-2004; = sectoral controls included. Robust standard errors. Complete estimations in Appendix 3.3.3. Source: own elaboration based on INDEC (2010).
Chapter 7: Innovative strategies, innovation persistence, and firm performance
- 226 -
7.4. Discussion of the findings
To return to the hypotheses, the positive relationship between past and present innovations
(H2.1) is rejected. This result is consistent with what was discussed in Chapter 3 in the
sense that in unstable environments the additional effect of past innovations is not verified.
Therefore, the literature about innovation persistence has to be applied with caution.
All hypotheses about innovation results and strategies (H2.2.1 to H2.2.3) are verified: the
probability of innovating depends positively on the innovative strategy. The ordering of the
probabilities confirms H2.2.4: firms with a continuous innovative strategy have higher
probabilities of innovating than the rest. However, H2.3 is not only rejected but the
opposite relationship is observed: the probability of innovating depends negatively of
having innovated in the previous period under a continuous innovative strategy. Firms with
a continuous innovative strategy show high probabilities of innovating but their past
innovations decrease the odds of achieving innovation in the present. The sign of the
coefficient among this group can be explained by the lock-in effect of innovation projects.
If these firms were developing a particular innovative strategy when the macro
environment and the rules of the game changed, then past innovation would no longer be
useful to gain a competitive advantage in the new scenario. In that case, they would have to
adapt to the new context, which would delay the achieving of positive results. However,
their better performance allows us to sustain that although firms with a continuous
innovative strategy had to overcome the lock-in effect of innovations, they were in a better
position to cope with the crisis and take advantage of the recovery and growth periods (the
unstable environment).
The new innovative firms, in contrast, started the innovation process looking for an
innovation suitable to the new environment, which would explain their higher probabilities
of achieving results than sporadic and non-innovative firms. These firms changed their
strategy together with the change in the environment, faced a lower lock-in effect—or no
effect at all—and could improve their performance afterwards.
With regard to sporadic innovative firms, although their innovative behavior and the lack
of impact on innovations resemble the characteristics of non-innovative firms, the better
results in terms of employment account for the impact of innovation on the firm’s
capabilities to deal with the selection process, even when it is an isolated event. Of course,
Chapter 7: Innovative strategies, innovation persistence, and firm performance
- 227 -
the fact that these firms achieve innovations more frequently than non-innovative firms
also provides evidence regarding the improved performance.
As far as the reviewed empirical evidence is concerned, the lack of confirmation of
innovation persistence obtained here partially agrees with those analyses where persistence
was confirmed only for some groups of firms. For Raymond et al. (2010), there is genuine
persistence only for companies belonging to medium-high and high technology sectors; for
Clausen et al. (2011), only when the innovation structures are based on science or market-
oriented; for Le Bas et al. (2011), when organizational innovations are achieved. In other
words, persistence is confirmed there where the analysis is conditioned to a particular type
of innovation process and not merely a serial correlation between events.
Regarding the disagreements, besides the limits of the persistence literature discussed in
Chapter 4, part of the explanation lies in the fact the we are comparing this study with
those based on firms that faced decades of macroeconomic stability (in Raymond et al.,
Dutch companies; in Peter German and in Clausen et al., Norwegian ones). Consequently,
to expect the innovative dynamic and innovation paths to fall within the same theoretical
developments or suggest the use of similar empirical approaches does not seem the most
accurate comparison. It is precisely in this respect that I argue that a more comprehensive
understanding and methodological treatment of the NIS surrounding the firm are required.
Another element of disagreement has to do with the definition of spurious persistence.
Results show that there are some characteristics of the innovative process that contribute
to innovation (innovation investments), but others contribute to it happening persistently
(the type of innovative strategy). In that sense, Raymond et al.’s (2010) spurious persistence
would be an innovative project which does not trigger more innovations, while true
persistence would be just another name for a virtuous innovative strategy sustained over
time. In any case, the results draw attention to the importance of understanding (and
controlling) the innovation process based on a dynamic approach in which firms changing
their strategy is allowed for.
H3.2 is another hypothesis that is not completely verified. Innovation results impact firm
performance positively in the case of exports and employment (H3.1.2 and H3.1.3) but do
not impact firm productivity. Once again, these results agree with the literature about the
Argentinean case and the low scope of the achieved innovations. At the same time, given
the large increase in sales levels (the denominator of the productivity ratio), these results
Chapter 7: Innovative strategies, innovation persistence, and firm performance
- 228 -
draw attention to the use of variables that are strongly affected by the economic cycle to
measure innovation. In this particular case, the increase in production levels after the crisis
was based on the incorporation of labor and the exploitation of the installed capacity (see
Chapter 2).
Finally, H3.2 is confirmed. Firms with a continuous innovative strategy have a better
economic performance than the rest. The analysis of the performance variables and the
impact of the strategies confirm the existence of different trajectories and, as with the main
conclusions of Chapter 6, the four innovative strategies correspond to four different
economic performances: continuous firms reached simultaneous improvements in
productivity, employment, and exports; new innovative firms managed to improve their
export and employment levels. In both groups of strategies, this positive impact was
verified regardless of the innovation results. Sporadic firms increased their employment
levels although gains in productivity and employment were subject to achieving innovation
results.
Graph 7.1 summarizes these findings in terms of the average level of annual growth for
each group. The X and Y axes show the increase in productivity and employment growth
levels (average annual growth rate), respectively; the circles account for the annual growth
in export intensity (exports to sales). To include sectoral controls, each firm’s values were
divided by the average for the sector they belong to, and then average values for each
strategy were calculated.
The group of firms with a continuous innovative strategy (C-C) presents more systemic
growth, with simultaneous gains in productivity, employment, and exports. Between 1998
and 2006, the productivity and growth (employment) of these firms increased around 5%
more than the sectoral average productivity and export levels to sales about 36%, in both
cases, per year. The new innovative (N-C) have a similar growth rate to the continuous
group (4%), although with lower gains in productivity and exports (3% and 12%,
respectively). Sporadic innovative firms present similar levels of productivity growth as new
innovative firms (2.7%) although with significantly lower levels of growth (1%) and
exports, which, in fact, grew below the average levels (-22%). Non-innovative firms, of
course, are below the average in all dimensions, with significant distances even from the
sporadic firms.
Chapter 7: Innovative strategies, innovation persistence, and firm performance
- 229 -
Graph 7.1: Average annual growth rate 1998-2006—Productivity, growth, and exports
Obs.: 800. Cont: C-C; New: N-C; NI: NI-NI; Spor: rest of the firms. Pd: labor productivity; Gr: employment growth; size of the circles: export intensity (values next to the strategy’s labels). In all cases estimated as average annual growth rate 1998–2006 with regard to each firm’s sectoral average, then the average values per strategy were calculated. Source: own elaboration based on INDEC (2010).
In terms of the literature on innovation and performance, the higher levels among
continuous and new innovative firms and the poor results among the rest agree with
Bottazzis et al.’s (2010) findings regarding the persistence of different productivity levels.
In fact, evidence shows that some low-productivity firms managed to survive the entire
cycle with a competitive strategy that was not based on innovation (non-innovative firms),
which means that low productivity firms persisted over time. The results also confirm the
positive relationship between innovation and performance (Arvanitis and Hollenstein,
1998; Kemp et al., 2003; de Jong and Marsili, 2006; Erbes et al., 2010; Yoguel et al., 2011;
Srholec and Verspagen, 2012), although they also show that this relationship is not directly
related to the innovation results but to the existence of different paths. In this sense, the
results agree with Percival and Cozzarin’s (2008) findings for Canadian firms, where the
authors find complementarities between innovation orientation, results, and performance,
and they also agree with Ito and Lechevalier’s (2010) findings for Japanese firms where a
highly innovative dynamic increases the odds of exporting but not the intensity.
In short, the results confirm the positive relationship between a sustained innovative
strategy and the firm’s performance. However, this better performance can be
acknowledged if the innovative dynamic is understood as a process and not merely as a
result. In this sense, our findings disagree with the literature which assumes a direct
relationship between innovation results and performance, and especially with Crépons et
Chapter 7: Innovative strategies, innovation persistence, and firm performance
- 230 -
al.’s (1998) and Yoguels et al.’s (2011) findings regarding the linear sequence between
innovation investments, innovation results, and productivity. It is not the persistence of the
results (the innovations), but the persistence of the innovative behavior that determines the
differential performance of firms. Continuous innovative firms reach simultaneous
improvements in the rate of innovation results, productivity, employment, and export
growth. New innovative firms managed to increase their probabilities of positive
innovation results with gains in productivity and exports but lower levels of employment
growth. Sporadic firms have higher levels of expenditure in some years, but these are not
sustained over time, and this impacts their economic performance by increasing their
employment levels but within a context of low productivity and exports. Only firms with a
sustained innovative strategy reach sustained innovation results and performance
improvements.
7.5. Preliminary conclusions
The objective of this chapter was to analyze the relationship between the innovative
strategies and the firm’s performance. Based on the results of the model and the findings
of the previous chapter, we can conclude that continuous firms seem to have followed a
successful high-profile innovative strategy, with simultaneous gains in productivity, growth,
and exports. These firms managed to improve their economic performance by reducing the
impact of the economic cycle and taking advantage of domestic demand and the export-led
growth given by the new environment. This took place within a context of high levels of
qualified human resources, investments, and linkages. The results coincide with the
literature in terms of the positive impact of innovation and shed light on the fact that this is
the direct result not of the achieved innovations but of a more dynamic behavior. As such,
the innovation process is an important factor to explaining the superior performance of
some firms, and the concept of the innovative strategy allows this to be captured.
A second group, made up of firms that changed their innovative strategy after the crisis,
also shows a more virtuous trajectory. These firms shifted from a non-innovative strategy
to an innovative one and, similar to the continuous group, increased their productivity
levels and grew. Therefore, these firms show path independence and demonstrate how
innovation decisions can explain path creation. Once again, the concept of innovative
strategies and the dynamic understanding of the innovation process allowed this
phenomenon to be examined.
Chapter 7: Innovative strategies, innovation persistence, and firm performance
- 231 -
From a methodological perspective, the results on new innovative firms provide evidence
regarding the importance of analyzing innovation in a dynamic way, where feedback and
accumulation processes are taken into account, allowing innovative strategies to be studied.
In traditional analyses, new innovative firms would have been classified as non-innovative
companies during the first sub-period, as middle-level innovative firms during the second,
and as dynamic innovative firms during the later ones. For each sub-period, average values
would have been the product of average levels of different behaviors (for instance, during
t-2 new innovative firms would have been together with the non-innovative ones), with the
consequent distortion of the average values, conclusions, and forecasts.
The sporadic group achieved employment gains, but in a context of low productivity and
integration into foreign markets. The innovative dynamic of these firms is associated with
isolated investments, without articulations with other dimensions of the firm. The results
show that this innovative strategy is successful in terms of growth, but not in terms of
productivity and exports. Once again, these are successful private strategies but they have
little impact on long-run firm growth.
In terms of the literature, these results confirm the impact of biased expenditures and the
importance of analyzing not only the level of investments but also their characteristics.
Traditional measures focus on the expenditures on capital goods and R&D and the impact
in terms on productivity. Based only on these measures, sporadic firms would have been
classified as highly dynamic innovative firms with high impacts on employment, while the
more complex approach presented here raises question regarding the impact of these short-
term strategies, based on high levels of isolated investments but bringing poor results in
terms of the quality of employment and the sustainability of a model of growth based on
reduced productivity and improvements in export levels.
To conclude, the evidence shows that some innovative strategies have been more
successful in the unstable Argentinean system than others. When looking at the continuous
innovative firms, it can be argued that innovation is a possible and profitable way of
competing even in unstable environments, which these firms learned to cope with.
However, the survival of non-innovative firms and the poor innovative dynamic of the
sporadic ones suggest that innovation is one of many ways of seeking a competitive
advantage. These firms also learned how to deal with the unstable environment. Finally, the
existence of firms with a new innovative strategy accounts for the possibility of path
independence and sheds light on how to promote path creation based on innovation.
- 232 -
Section IV: Conclusions
8.1. Research questions and findings
This thesis has analyzed the relationship between the NIS and heterogeneity at the firm
level. The main research question was about the existence of firms better prepared to deal
with unstable environments, with the objective of studying how innovative strategies
change over time. We claim that the answer to this question lies in the combination of two
sources of micro-heterogeneity, both shaped by the national environment but ultimately
determined by the firms’ learning capabilities.
One source of micro-heterogeneity arises from the institutional set-up affecting the
selection process. Besides Schumpeterian creative destruction, the NIS impacts the
organizations that rule market competition by imprinting structural features on them,
which are later combined with the institutional culture and routines. Their specific dynamic
and goals impact the selection process by limiting the extent of Schumpeterian creative
destruction, to the point that firms with different levels of productivity are able to remain
in the market. Firms face this context-specific selection process in different ways,
sometimes based on innovation, and sometimes not based on innovation at all. As a result,
the productive structure will be made up of a heterogeneous set of agents, which interact
with a heterogeneous set of organizations, which in turn affect the competitive dynamics.
The second source of micro-heterogeneity arises from the firm’s learning capabilities. As a
result of the environment, firms follow different competitive strategies. The strategy is the
set of decisions regarding how to face the selection process and it results from the firm’s
interpretation of the environment (and the NIS) and its resources and capabilities (path
dependence). Since firms can learn, they are able to re-interpret the environment and to
adjust the allocation and exploitation of their resources, that is to say, to adjust their
strategy. Given the multiple possible combinations of decisions, capabilities, and resources,
and the multiple possible path dependences, each firm will be different. Once again, the
system will be made up of a heterogeneous set of firms, this time because of the multiple
possible results of the competitive strategies.
Section IV: Conclusions
- 233 -
The NISs of unstable environments are characterized by a weak institutional set-up, where
the incentives to compete based on innovation are limited. Therefore, the resulting first
type of micro-heterogeneity will be deeper and the range of extra-innovation opportunities
for sustaining a competitive advantage will be larger. However, the instability of the system
will also be part of the firm’s path dependence and past learning processes, and will have
imprinted specific capabilities to deal with this unstable system on the firm. Consequently,
firms with higher capabilities will be better prepared to take advantage of environmental
opportunities and to avoid environmental risks. In this way, they are able to understand the
environment and foresee how resources could be better allocated and exploited. Within
this scenario, although innovation is not the only way of competing, it is the competitive
strategy that has the most impact on capabilities. Since the innovation process leads to
knowledge creation and appropriation, it triggers additional feedback and accumulation
processes onto the firm’s capabilities and resources. As a consequence, those firms with a
strategy based on innovation will be better prepared to deal with unstable environments.
The empirical work was based on a group (800) of Argentinean manufacturing firms for a
particular period (1998–2006). As a consequence, these conclusions are preliminary, and
only the reproduction of this approach and analysis of other environments or periods will
allow the results included here to be extrapolated so as to reach a more generalized
understanding of the relationship between the NIS and heterogeneity at the firm level.
However, the preliminary nature of these conclusions does not undermine their usefulness
in terms of improving understandings of the innovation process within the firm and how
the NIS and the environment in general impact decisions at the micro level. In this respect,
the findings of this thesis have theoretical, methodological, and practical implications,
associated with the specific objectives (SO) that guided the research project, namely:
SO1: To make a contribution to the NIS approach in terms of how it deals with the
existence of micro-heterogeneity in unstable environments;
SO2: To identify and characterize different innovative strategies and their different
impacts on productivity, labor dynamics, and exports;
SO3: To make a contribution to the understanding of how innovation strategies work,
in order to propose key criteria for public policies capable of fostering those more
successful innovative behaviors.
Section IV: Conclusions
- 234 -
The theoretical implications are related to the interaction between the NIS and the firm’s
strategy (SO1), including the impact of path dependence but also the sources of path
creation (the re-interpretation of the environment). The methodological implications arise
from the need for a dynamic analysis of the firm’s strategy in order to capture the learning
processes which determine it (SO2). The practical implications are associated with the
policy recommendations and the possibility of extending the results to other environments
(SO3). In the next section, these implications are further discussed. Following that is a
discussion of future research lines opened up by the limitations of this thesis.
8.2. Theoretical implications
8.2.1. The NIS and micro-heterogeneity
To accomplish SO1, the NIS approach and subsequent contributions to it were analyzed
and then applied to the Argentinean case, in order to study how it deals with the
characteristics of an unstable environment. In conclusion, it was observed that the
existence of micro-heterogeneity is not fully acknowledged by the approach and the
possibility of finding firms that escape the system’s lack of development is ruled out from
the analysis (Chapters 1 and 2).
According to the NIS approach, the research question of this thesis would have been
answered by explaining the existence of outliers—successful cases, or firms that managed
to escape from the structural causes that determine a generalized poor innovative
performance. However, this research has shown that these firms are not outliers but part of
the system, moreover, they are endogenous to its dynamic. This means that the same NIS
that leads to the existence of firms with poor innovative dynamics also leads to the
emergence of firms with high-profile innovative behavior.
In terms of the literature, the same systems that create Dutrenit and Katz’s (2005) stylized
facts about the poor innovation process in Latin America lead to the existence of
successful innovative firms in the region. This implies extending Freeman’s (1995)
historical determinants of firms’ behavior to understand how they have learnt to face the
selection process, whether by means of a higher performance or by escaping
Schumpeterian creative destruction. Historical processes have shaped firms’ behavior and
some of them capitalized on this history in the form of capabilities for understanding the
environment, foreseeing the short-term, and taking advantage of opportunities (e.g. a single
firm cannot influence the exchange rate but it can learn to be flexible enough to import
Section IV: Conclusions
- 235 -
goods when it goes down and to quickly substitute them when it goes up). This scenario
highlights the limits of the two types of studies within the NIS literature, known as
structural analysis and the distance-based approach.
What I have argued is that the problem with those more structural approaches (Dutrenit
and Katz, 2005; Katz, 2007; Anlló et al., 2007; Porta, 2006, Bisang, 1995; among others) is
that they do not pay sufficient attention to firms’ learning capabilities and how some of
them learned how to make extraordinary profits based on innovation. In this respect, I
claim that there is a need to look at “successful” cases in order to understand why these
firms, subject to the same structural incentives, the same history, and the same unstable
environments, present behaviors and results that escape the theoretical predictions.
Regarding the distance-based literature (Alburquerque, 1999; CEPAL, 2008; Cimoli, 2005;
Godinho et al., 2004; Nelson and Dahlman, 1991; OECD, 2007, among others), the
existence of micro-heterogeneity implies that development is not just a matter of distances.
The problem with these explanations is that they neglect the impact of the system on the
institutions affecting the selection process and how firms understand these incentives. In
this sense, it is not a matter of closing the gap between the inputs of the innovative process
or the need for a specific approach to development. It is a matter of “catching up” with
innovative behaviors (e.g. to foster more investments on R&D would not improve the
competence building of the system unless firms have the willingness to pursue an
innovation-based strategy and the capabilities to implement it).
In conclusion, I propose to go back to the origins of the approach, to Lundvall’s (1992)
focusing device, in order to understand the process of competence building, but accepting
the fact that it is an evolving system made up of heterogeneous organizations and complex
interactions. The results of this dissertation prove that looking at the system is a useful tool
for approaching the complexity of technological development, but not by means of making
an inventory of the organizations affecting the innovation process—Lundvall’s (1992)
narrow and wider system—or measuring levels of inputs and outputs—Nelson’s and
Dahlman’s (1991) approach (among many others). The concept of the NIS is a
methodological tool that allows the identification of those organizations that, within a
given moment and within a particular environment, affect the behavior of the
organizations and individuals that determine the capitalist process of production,
distribution, and consumption. As such, to jump from these national systemic determinants
to how they impact firms’ innovative behavior, a deeper understanding of the process of
Section IV: Conclusions
- 236 -
competence building at the micro-level is required. This is a missing “building block”
within the approach and an intended contribution of this thesis.
8.2.2. The NIS and innovative firm
The literature on the innovative process at the firm level provides solid arguments
regarding the sources of micro-heterogeneity and how the combination of resources,
capabilities, and knowledge determine how the firm faces the selection process, given its
environment and its structural determinants (Nelson and Winter, 1982; Penrose, 1959;
Lundvall, 1992; Metcalfe, 2002; Malerba, 2004). This literature can also explain why firms
with different levels of productivity remain in the market, how path dependence affects the
range of decisions firms can take, and how learning is at the core of the firm’s innovation
process (Bottazzi et al., 2010; Antonelli, 1997; Lundvall and Lam, 2007). However, the lack
of attention to impact of the firms’ learning capabilities on its path of growth worked
against the explanations regarding the nature of the firm’s competitive dynamic, the
possibility of a change in the firm’s strategy, and how it can alter its path dependence.
In this respect, I argue that there is a need to understand how learning processes and
interactions with the system affect the firm’s competitive strategy (Chapters 3 and 4). The
continuous interaction between the strategy, the structure (resources and capabilities), and
the firm’s environment triggers learning processes within the firm. Depending on the firm’s
capabilities, it will adjust the strategy, which is now the result of higher levels of resources
and enhanced capabilities (learning-by-doing, -using, -interacting, etc.). Even if the firm
fails in its strategy and there is no increase in resources, the learning processes have
enhanced its capabilities to deal with the environment (learning-by-mistaking). Therefore,
whether the firm has made profits from the selection process or not, from one moment to
the following one, it will adjust and even radically change its strategy. As a result, the
concept of strategy contributes to the understanding of the firm’s path dependence, the
sources of path creation, and the impact of the interaction between the firm and the NIS.
From a theoretical perspective, if firms follow strategies determined by their understanding
of the environment, its resources, and its capabilities, then there is no reason to believe that
there is an average response to environmental incentives (the structural approach), nor
similar distances between the most innovative and least innovative firms (the distance-
based approach). In this respect, our findings provide evidence regarding the existence of
different strategies, which are present in firms which have faced the same environmental
Section IV: Conclusions
- 237 -
incentives and structural determinants (e.g. sector and size). That is, the existence of micro-
heterogeneity. Moreover, since firms tend to adjust their strategies, micro-heterogeneity
tends to changes over time.
With regard to path dependence and path creation, I argue that the firm’s current
competitive position is the result of past strategies, such that the literature on path
dependence and structural determinants only partially applies. Since the strategy is the
result of current resources and capabilities, it is path dependence. However, to the extent
that firms can learn and improve their capabilities, they can also alter their strategy, escape
from their path dependence, and get into a new path of growth. My findings show that
there are signs of path independence, which are explained by the level of internal and
external capabilities (human resources and linkages), which were later combined with new
resources and capabilities (investments in knowledge) and led some firms to change their
competitive strategy and succeed in the market (innovation results and better
performance). My results provide evidence regarding the usefulness of a strategy-based
approach to understanding how path dependence and path creation interact. Consequently,
looking at the firm’s strategy also allows us to understand how feedbacks and accumulation
processes interacts with firms’ learning capabilities, allowing accumulation.
At the aggregate level, that is, the NIS, the strategy-based approach can contribute to the
understanding of why successful firms can be found in unsuccessful environments, and
why micro-heterogeneity is shaped by the NIS but it is ultimately determined by the firms’
strategic behavior. The firm’s interpretation of environmental signals (incentives and
disincentives) is mediated by its capabilities. As such, they will determine the extent to
which the firm is able to take advantage of environmental opportunities or to avoid
environmental constraints. Since firms can learn, they can improve their comprehension of
the environment, enhance their capabilities, and manage to grow and survive regardless of
the environment. Of course, the more adverse the environment, the higher the capabilities
necessary to survive, while the lower the incentives required to follow an innovative
strategy. It is precisely in terms of this apparent contradiction that my conclusions
regarding how firms deal with unstable environments can be found.
8.2.3. Innovative strategies and unstable environments
According to the literature, unstable environments generate disincentives to innovate
(Chapter 1). First, because the level of macroeconomic uncertainty increases the risk of any
Section IV: Conclusions
- 238 -
investment project (Kosacoff y Ramos, 2006). Second, because the weak selection process
opens up the possibility of competing and surviving without innovation (Bottazzi et al.,
2010). Third, because the changes in the rules of the game create incentives to look for
short-term strategies regardless the impact they can have on the long-run growth
(Fajnzylber, 1989). My findings show that although these are good explanations for some
firms’ behavior, they cannot explain the existence of firms with a high-profile innovative
strategy in unstable environments.
Our results show that some firms continuously invested in innovation, others did so
sporadically, others did not invest in innovation at all, and others shifted from non-
innovative behavior towards a high-profile innovative strategy (Chapters 5, 6, and 7). All
these firms were part of the same environment and faced the same (dis)incentives
emanating from Argentina’s innovation system. Given these results, existent literature can
explain the case of non-innovative and sporadic innovative firms, based on the short-term
incentives and the macroeconomic disincentives. It can also explain why some firms
present low levels of investments in innovation, biased towards the external acquisition of
knowledge (capital goods) with low levels of interaction with the NIS (Dutrenit and Katz,
2005; Anlló et al., 2007). However, it cannot explain the sustained behavior of continuous
innovative firms, the change in the strategy among new innovative firms, and the existence
of firms with high levels of investments in innovation, which combine internal with
external knowledge creation and appropriation.
If answers are sought within the literature on the heterogeneity of innovative behaviors at
the firm level, the findings of this research would have been explained as a matter of
adaptive responses to the signals from the environment. More specifically, they would have
been analyzed in terms of knowledge-making vs. buying decisions (e.g. Arundel and
Hollanders 2005; Nelson and Dahlman, 1991; Fraga et al., 2008), subject to
macroeconomic evolution (e.g. Kosacoff and Ramos, 2006; Dutrenit, Katz, 2005; Ocampo,
2012). However, my findings provide evidence regarding a more complex dynamic than a
specific innovative behavior—Srholec and Verspagen’s (2012) innovation ingredients—or
a sector-specific source of technological change—the application of Pavitt’s taxonomies
(e.g. Clausen, et al., 2011). The results show continuities and discontinuities, firms which
persisted with the same behavior when the environment changed, and firms which decided
to change it. The results also show how Lundvall’s DIU and STI modes of learning are
combined not only at the firm level but also over time (Jensen et al., 2007), and how
Section IV: Conclusions
- 239 -
Lugones’ et al. (2007) balanced firms are attached to long-term innovative behaviors. In
other words, the results show a dynamic process of strategic decision, implementation, and
adjustment which leads not only to changes in the trajectory of the firm but also in the
characteristics of the micro-heterogeneity.
In short, the more macro approaches to firm behavior fail to predict the different micro-
responses to environmental incentives. The more micro approaches, in turn, fail to predict
how feedbacks and accumulation processes shape the interaction between the firm and the
environment. Once again, thinking of the firm as an organization that takes strategic
choices and learns can shed light on both matters, especially in the case of firms from
unstable environments.
Firms within unstable environments do not act as if stability is about to come. Either they
learn how to compete in this environment or they disappear from the market. An unstable
environment such as Argentina is not the result of an external shock, an exogenous
disruption (such as the ones referred to in the resilience literature). Instability is an
endogenous feature given by the aggregation of multiple characteristics of the system (the
productive structure, the political system, the national idiosyncrasy, the place it has in the
international context, etc.). Then, the same way we can expect disincentives to innovate, we
can expect learning processes associated with this particular unstable environment.
If instability is a regular feature, then firms could have learnt to compete and survive in this
environment. Some of them will choose to seek short-term profits with low investments in
innovation (what the literature predicts). Others, however, will try to base their competitive
advantage on the search for continuous productivity improvements (what these findings
suggest) to avoid the external competition, to diminish the impact of the downturns in the
national income or to search for external markets to relax the fluctuations given by the
domestic economic cycle. Of course, the latter are strategies that demand higher
capabilities (the case of continuous and new innovative firms). However, these higher
capabilities also increase the odds of carrying on a successful strategy (the performance of
these strategies) and determine more complex learning processes that feedback the already
high capabilities. In other words, these firms provide evidence regarding the possibility of a
high-profile innovative strategy in an unstable system. Learning, feedbacks, and capabilities
are at the core of the explanation about their existence and the firm’s strategy is what
allows us to look at how these processes interact.
Section IV: Conclusions
- 240 -
The other side of the story is the existence of firms with low capabilities. These firms could
be trapped in a low-profile innovative strategy, with great difficulties in enhancing their
capabilities sufficiently to look for a more successful strategy. Sporadic and non-innovative
firms (as well as those that disappeared from the market) could account for this possibility.
However, new innovative firms provide evidence regarding the possibility of breaking this
vicious circle and show how learning, accumulation, and capabilities are key to seeking a
more successful innovative strategy, even in unstable environments and even within a poor
innovative path dependence. In this respect, capabilities and learning processes are key to
understanding the sources of path creation, and a dynamic approach to strategy is the
methodological way of acknowledging this.
To conclude, our findings show that continuous innovative firms seem to have learnt how
to avoid the disincentives to innovate that emanate from the environment. New innovative
firms may have chosen to follow a similar path by means of pursuing a high-profile
innovative strategy. The existence of non-innovative and sporadic firms proves that
innovation is not the only way of surviving, although said groups performed worse. As
such, strategy, capabilities, and the learning processes are key to understanding how some
firms are better prepared to deal with unstable environments.
8.3. Methodological implications
The second specific objective is linked to the methodological implications of the findings
of this thesis. Given the objectives, environmental instability was contrasted with micro-
regularities. The research question was split into three methodological questions, which
were later transformed into testable hypotheses:
Q1: Is there a positive relationship between the continuity of the innovative strategy and
the complexity of the innovative structure? (Chapter 6)
H1.1: Continuous innovative firms have a more complex innovative strategy.
H1.2: New innovative firms changed their innovative strategy in the direction of a more
complex structure over time.
H1.3: Sporadic innovative firms have a low-complexity innovative structure sustained over
time.
Q2: Is there is a positive relationship between the continuity of the firm’s strategy and its
past and present innovation results? (Chapter 7)
Section IV: Conclusions
- 241 -
H2.1: The probability of innovating depends positively on having innovated in the past.
H2.2: The probability of innovating depends positively on the innovative strategy.
H2.3: The probability of innovating depends positively on having innovated in the
previous period when following a continuous innovative strategy.
Q3: Is there is a positive relationship between the continuity of the firm’s strategy, its
innovation results, and economic performance? (Chapter 7)
H3.1: Innovation results have a positive impact on firm performance (productivity, labor
growth and exports).
H3.2: Continuous innovative firms improve their performance improvements, regardless
of their innovation results.
To test the hypotheses I constructed a model which was later applied to a set of 800
Argentinean manufacturer firms for an unstable period (1998–2006). The model was based
on the three main streams of literature on innovation at the firm level: the literature on
micro-heterogeneity, innovation persistence studies, and analyses aimed at testing the
relationship between innovation and performance.
In this case, the contribution to the literature lies in the inclusion of the dynamic nature of
firm behavior and the possibility of a change in the firm’s strategy as a proxy for learning
capabilities. The working hypotheses aimed to analyze continuities and discontinuities in
the firm’s innovative investments attached to specific innovative structures, as a proxy for
the innovative strategy as well as a way of testing changes in the characteristics of micro-
heterogeneity. In general terms, the findings allowed the objective (describing the
strategies) to be fulfilled and the main hypothesis (the existence of changing micro-
heterogeneity and firms better prepared to deal with the environment) to be confirmed.
These results have different implications regarding the existent literature.
8.3.1. Innovative strategies and micro-heterogeneity
Regarding the literature on micro-heterogeneity, our findings (Chapter 6) confirm that
some firms maintained complex innovative strategies (confirming H1.1) while others
changed their innovative behavior from non-innovative to highly innovative (thus
confirming H1.2). However, some firms also changed in the opposite direction (from an
innovative to a non-innovative strategy) and present just isolated investments in innovation
attached to a low-complexity strategy (confirming H1.3). Therefore, firms change from one
Section IV: Conclusions
- 242 -
cluster to another and from one type of knowledge investment to another (buy vs. make).
Sometimes this responds to a change in the strategic way of facing the market, sometimes
this is just the result of the expected process of innovation creation, exploitation, and
replacement (Utterback and Abernathy, 1975; Nelson, 1994). This research cannot account
for the specific causes behind the change in the strategy, however, it provides evidence
regarding the importance of accounting for different strategies and the fact that static
(cross-sectional) analysis of micro-heterogeneity accounts only for a partial time-specific
aspect of the innovative process, with low predictive capacity as to how it will evolve,
which part of the innovative process the firms are investing in, or the factors that allow
firms to sustain a complex innovative strategy. In other words, existent literature can
describe a specific innovative structure at the firm level (a snapshot) but it cannot fully
characterize the firms’ innovative strategy. My conceptual approach to the firm’s strategy
proposes a methodological tool to overcome these limitations.
Another set of findings that could improve the characterization of firm behavior arises
from the continuity criterion as a source of micro-heterogeneity. The results show that the
inclusion of the frequency of expenditure could improve estimations by reducing the
unobservable effects and controlling the different characteristics of investments in
innovation (continuous R&D investments are different from sporadic acquisitions of
capital goods, even when they represent similar sums of money). However, the results also
showed that the frequency criterion does not replace the traditional structural variables
(size, sector, capital ownership, etc.), but rather complements them, as they are key to
avoiding the bias given by the technological regime. Furthermore, these structural variables
help to control the differential impact of sectoral/size-related public policy, the existence
of a specific knowledge infrastructure, and the country’s productive history.
To conclude, results show that a dynamic approach to firm behavior which includes the
possibility of a change in the strategy allows innovative behavior to be better understood.
This in turn sheds light on the possibility of different responses to the same changes in the
environment and provides elements through which to characterize the knowledge creation,
application, and appropriation processes.
8.3.2. Results about the persistence of innovation
The first conclusion from the persistence analysis (Chapter 7) is the lack of confirmationof
the correlation between past and present innovation results (H2.1. is rejected). The results
Section IV: Conclusions
- 243 -
differ from the situation that had been predicted on the basis of theoretical arguments and
other empirical findings. The second conclusion was that those firms with a more coherent
and sustained innovative strategy have higher probabilities of achieving innovation results
(H2.2 is confirmed for continuous firms) but with a negative innovation persistence (H2.3
is rejected). The interpretation of the higher probabilities is quite tautological: firms with a
more sustained quest for innovations persist in innovating. However, when the innovative
structure is controlled, the results show that sustained investments increase the odds of
achieving results, regardless of the actual level of investments. With regard to the negative
impact of past results, the explanation may lie in the lock-in effect of past innovation
within a context of changes in the environment: in the rules of the game, in relative prices,
in the economic cycle, etc.
The results also showed higher probabilities among firms with a new innovative strategy in
comparison with non-innovative firms, while there are no significant differences in the
probabilities of innovating between these last two groups (H2.2 is verified for new
innovative firms but rejected for sporadic innovative firms). The results lead one to wonder
about the applicability of persistence arguments to unstable environments and the use of
average lags for all firms when we know for a fact that different investments will have
different return periods.
From the methodological point of view, the frequency criterion was used to identify the
strategies, and the results show that firms with different strategies face different
probabilities of persistence (or no probabilities at all). These results were based on the
inclusion of a multiplicative term affecting past innovation in addition to the dummy
variables to control initial levels of each innovative strategy (as common practice). In
economic terms, this means that firms with different strategies (different ways of achieving
innovations) face different innovative trajectories, with different paths of marginal
improvements, types of innovation, and degrees of novelty. Therefore, the impact of past
innovations is not necessarily captured by the present results, so persistence-related
assumptions have to be applied with caution.
This leads us to the third conclusion of this analysis: there exist different strategies with
different impacts in terms of innovation, but there is no clear relationship between the
innovation investment level and past and present innovation results. In some cases, this is
because past innovations are the result of sporadic isolated events, in other cases, it is
because past innovations are no longer useful for dealing with the new environment; in
Section IV: Conclusions
- 244 -
others the lag between one innovation and the following one changed because the type of
innovation has also changed. Then there are multiple explanations about the lack of
persistence, which accounts for the need to examine firm performance in more depth if we
want to understand the relationship between innovation investments and results.
8.3.3. Innovation process, results, and impacts
The last set of methodological implications is related to the impact of innovation on
performance (Chapter 7). The findings confirm the positive impact of the innovation
results on the firm’s performance in the case of employment growth and exports, but they
do not impact productivity levels (H3.1 is partially verified). The findings also show that
the innovation process among continuous innovative firms impacts firm performance,
regardless of the innovation results (thus confirming H3.2). With regard to the literature,
the results agree with the already established positive impact of innovation on growth but
shows that there is no linear relationship between the innovation process, innovation
results, and firm performance, in contrast to the conclusions of Crépon et al. (1998) and
Yoguel et al. (2011).
This analysis supports the claim that firms with an innovation-based strategy perform
better, which could be explained by their higher capabilities and the learning processes that
arise from investments in knowledge, which are not necessarily mediated by the innovation
results. This means that looking at the innovative strategy is a good way of predicting the
impact of innovation. Conversely, it also means that innovation results are not always a
good proxy for the impact of innovation on performance.
Finally, the use of different measures of performance (productivity, labor, and exports)
showed that different innovative processes will affect the firm’s relationship with (and
impact on) the environment in different ways: sometimes innovation leads to higher
productivity levels combined with an increase in the quality and quantity of employment,
sometimes to higher employment levels but low skills levels, and sometimes to a higher
export intensity but low levels of innovation investment. Therefore, although innovation is
always good for the firm, not all innovations are equally good for the rest of the system.
These results lead us to the final set of implications: the possibility of extending these
results to other countries and the policy recommendations that can be extracted from this
research, which constitute the last specific objective of this thesis.
Section IV: Conclusions
- 245 -
8.4. Practical implications
8.4.1. Application to other countries
This research project began by analyzing the IS approach and how it can be applied at
different levels (national, local, sectorial). It then went on to explain the impact of the
national level and how the national organizations shaping competition and the learning
processes at the firm level determine the existence and characteristics of micro-
heterogeneity. This led to the quantification of the relevance of the national approximation
to account for the innovation dynamic. The arguments and hypotheses were tested for the
case of Argentina—the obvious question is to what extent the results can be applied to
other national contexts.
The first distinction made was between developed and developing countries (Chapter 1). In
this case, it is argued that this classification is not relevant to understanding the impact of
the national system on the organizations shaping competition and how firms learn to deal
with the environment. My explanation of micro-heterogeneity applies to both developed
and developing countries. Of course, country-specific analyses should answer the question
about how the NIS has impacted the selection process, firms’ learning capabilities, the
accumulation of skills, and access to resources. In other words, how a particular NIS has
shaped national micro-heterogeneity.
The second distinction made was between Latin America and the rest of the world. In this
case, the region has some features that distinguish it from developed countries but also
from the least developed ones. These features are captured by the NIS literature without
the need for a specific approach for development. Our analysis of the NIS and the issue of
development was focused on Latin America, and a deeper understanding of the literature
on other less developed regions would allow the analysis to be extrapolated to other types
of developmental paths. However, this does not undermine the possibility of the strategy-
based approach and explanations of the sources of micro-heterogeneity being applied to
other regions, where other histories and development levels should be taken into account.
A third distinction was made between Argentina and the rest of Latin America. The focus
on Argentina was based on its history of instability and changes in the rules of the game
(Chapter 2). Although there are some features that distinguish Argentina from the rest of
the region (the specific natural resources, the early implementation of labor regulations and
universal education, and the nature of its populist parties), similar macroeconomic
Section IV: Conclusions
- 246 -
instability can be observed in many other countries in the region (such as Brazil, Mexico,
Venezuela, and Uruguay). Moreover, the 1982 and 1995 crisis started in Mexico; during the
1990s, Brazil had a similar convertibility plan which ended with a short but deep recession
in 1998; in the same decade the Washington Consensus impacted most South American
countries in the form of deregulation and liberalization; and in what has gone by of the 21st
century, a new wave of left-leaning governments have come to power in the Southern
Cone. Thus, although the characterization of the national process of competence building
implies the need for specific country-based studies, the analysis of the impact of instability
on the innovation process and the relevance of the study of firms better prepared to deal
with an unstable environment also applies to other Latin American countries.
Finally, there is the continuity criterion to account for the different strategies and how it
can help to identify different innovative strategies in other countries. Despite the potential
sources of region-specific characteristics, results from the theoretical discussion allow us to
sustain that firms with a continuous strategy should face additional learning processes
which should impact firm performance (Chapter 3). Consequently, from one moment to
another, firms can continue or modify their behavior and this study has provided evidence
to sustain that this is relevant to understanding how firms deal with the the environment,
regardless of how developed the country is (Chapters 6 and 7). Of course, how continuity
is translated into a strategy (e.g. the span of years), what the best proxies for the innovative
structure are, and how capabilities should be approximated are all aspects that will depend
on the specific research questions, data availability, and the characteristics of the country.
8.4.2. Policy recommendations
As was mentioned before, this study provided some evidence for the Argentinean case
based on a reduced group of firms for a specific period. Therefore, the policy implications
are preliminary, subject to future research on the issue of firms’ innovative strategies.
The first preliminary policy implication of this research is that there is no macroeconomic
situation that automatically triggers innovative strategies in firms. Our results show that
non-innovative firms are present during recession, recovery, and the growth period
(Chapters 6 and 7). This is consistent with the theoretical arguments about the causes of
micro-heterogeneity. During the growth periods, firms can survive with lower productivity
levels, given the expansion of domestic demand, or they can use additional profits to invest
in innovation. During recessive periods, firms can survive based on less sustainable
Section IV: Conclusions
- 247 -
practices (e.g. avoiding environmental regulations), downsizing, or minimizing profits, or
they can take higher risks and look for an innovative way to survive. Innovation policies are
required regardless of the moment in the economic cycle.
The second preliminary policy implication has to do with the impact of the NIS on the
existence of micro-heterogeneity. In this case, given the impact of the incentives generated
by other organizations, one key element of the innovation policy is its interaction with
other national, regional, and sectorial policies. This research has proved that innovation is
not a matter of isolated behaviors but the result of an interactive process between the firm
and the environment. Analogously, innovation policies should interact with other policies
in order to maximize their impact. Hence, innovation policies should be part of a wider
context of public policies aimed at fostering technological change and development.
A third preliminary policy implication is related to the difference between the innovative
strategy and the innovation results, at least for the case of Argentina. This research has
proved that some strategies are better than others in terms of their impact on the
environment (productivity, employment, and exports). It has also shown that this improved
impact is not fully captured by the innovation results. The differential impact is associated
with different (and sustained) innovative behavior. This means that looking at what specific
innovative strategy a public policy has induced is more important than counting the
number of innovations it has achieved. It also means that looking at the type of innovation
investments and how they evolve over time (e.g. the balance and continuity of expenditure)
is as important as measuring the relative level of expenditure to sales (or R&D to GDP).
The last preliminary policy implication refers to the role of capabilities. As was
demonstrated, the basis of the more successful dynamics can be found in the firm’s
capabilities. From a theoretical perspective, this is because they allow the firm to learn.
From an empirical perspective, because the level of qualified human resources was a key
difference within continuous and new innovative firms. Therefore, innovation is not only
about R&D and machinery, it is about human resources and how firms invest on creating
and enhancing organizational, productive and innovative capabilities. If we want to foster
innovation we should look at the capabilities the firm has—and those it lacks—to design,
implement, and adjust a successful innovative strategy.
Section IV: Conclusions
- 248 -
8.5. Limitations and future lines of work
The objective of this final section is to discuss the limitations of the present research, some
of which are associated with the need to narrow down the research question and select a
theoretical framework (Chapters 1 to 4), and some of which are explained by data
availability to analyze a heterogeneous set of firms over time (Chapters 5 to 7). However,
most of these limitations can be overcome with future research and, especially, by
interacting with other branches of innovation studies and other sciences.
Used as a theoretical framework, the NIS approach is a tool for studying one dimension of
the development process at the national level. When applied to a specific country, it can
help explain the process of competence building but it is not enough to explain the process
of structural change, the dynamic of the education system, the nature of the political arena,
or the implications and requisites for social development. In this respect, the NIS approach
can help to identify innovation’s contribution to the process of economic growth and
increasing welfare, but it is just one out of many causes of the complex process of
development. Establishing links between this approach and other theories that are also
concerned about inclusive development constitutes a promising line for future research,
especially in the field of social innovation, higher education, and productive systems.
Another interesting link would be with sectorial studies and economic geography, which
have long analyzed the impact of economic activities and regions on firms’ innovation
processes and performance. The characteristics of specific sectorial systems of innovation
and regional resilience are two elements that could contribute to explaining the firm’s
innovative strategy and the existence of differential incentives to innovate, given the
environment. In this respect, other branches of innovation studies (work organization,
regional studies, technological trajectories) and future studies about competence building
could improve understandings of the learning process within the firm, how path
dependence affects its strategy, and what the specific sources of path creation are.
A third interesting line for inter-disciplinary research arises from the definition of strategy.
The present research did not examine the determinants of strategy in depth, nor why a firm
decides to change it. Similarly, it has not investigated the sources of reductions in firms’
capabilities or the impact of different types of interactions with the environment. In this
respect, the management literature has long studied the process of decision-making, the
implementation and adjustment of strategy, and the ways a firm can enhance or lose
Section IV: Conclusions
- 249 -
capabilities (e.g. loss of employees). Besides future research based on evolutionary
approaches to the firm, building bridges with management literature could help understand
the competence building processes within the firm and how firms select a specific strategy.
The fourth limitation arises from the treatment of the relationship between the firm and
the environment. In this respect, complexity theory has dedicated much attention to the
knowledge flows within systems and between the firm and a network of agents. This
stream of evolutionary economics is also focused on overcoming the limitations inherent in
the static and linear approach’s measurement of interactions as linkages or cooperation
agreements. Building bridges between the NIS approach and complexity theory could help
us better understand and characterize how the environment impacts micro-heterogeneity.
The fifth set of limitations is given by the data used in this thesis. In this respect, the results
cannot be extrapolated to the entire Argentinean manufacturing population, still less to the
Argentinean productive structure. At the same time, the existence of firms that followed a
continuous innovative strategy and failed (by disappearing from the market) cannot be
studied. However, the data used allowed different strategies to be identified, different
innovative paths to be traced, and the importance of firm-level databases to be shown. I
am confident that future rounds of innovation surveys, other firm-level exercises, and the
improvement of the Argentinean statistical system, will allow researchers examine the
characterization of the firm’s innovative strategies and performance in greater depth.
To conclude, and despite these limitations, Argentinean history has shown that pure
market competition is not enough to trigger successful innovative strategies and also has
noxious impacts on society. Protectionism has not been enough either, and improvements
in welfare have come together with restraints on Argentina’s foreign and public accounts.
In a wider setting, the recent global financial crisis and the deep recession in the south of
Europe have proved that development is not just about increasing levels of R&D, the
participation of dynamic industries, or patent numbers. All this recent history provides
evidence for the need to challenge the established assumptions and to improve our
understanding of the role of innovation within the complex process of growth and
development. In this sense, the limitations of this research are another way of saying that to
understand a complex reality, analysts have to be able to design a complex investigative
approach. I hope this thesis has contributed in that direction, at least by shedding some
light on key features for a future research agenda.
- 250 -
Bibliography
Alburquerque, E., 1999. National Systems of Innovation and Non-OECD Countries: Notes About a Rudimentary and Tentative “Typology”. Brazilian Journal of Political Economy 19, 35-52.
Altimir, O., Beccaria, L., 1999. El mercado de trabajo bajo el nuevo régimen económico en argentina. CEPAL, Serie Reformas Económicas 28.
Andersen, E.S., 1992. Integration, Innovation and Evolution, in: Lundvall, B.Å. (Ed.), National Systems of Innovation. Towards a Theory of Innovation and Interactive Learning. Pinter Publishers, London.
Anderson, T.W., 2003. An Introduction to Multivariate Statistical Analysis. 3rd ed. Wiley, New York.
Anlló, G., Lugones, G., Peirano, F., 2007. La innovación en la argentina postdevaluación, antecedentes previos y tendencias a futuro, in: Kosacoff, B.e. (Ed.), Crisis, recuperación y nuevos dilemas. La economía argentina 2002-2007. CEPAL, Santiago de Chile, Buenos Aires.
Anlló, G., Suárez, D., 2008. Innovación: algo más que I+D. Evidencias Iberoamericanas a partir de las encuestas de innovación: construyendo las estrategias empresarias competitivas, in: RICyT (Ed.), El Estado de la Ciencia. Principales Indicadores de Ciencia y Tecnología Iberoamericanos/ Interamericanos. RICyT, Buenos Aires.
Ansoff, H.I., 1973. Toward a strategic theory of the firm, in: Ansoff, H.I.O. (Ed.), Business strategy. Penguin Books, Westminster.
Antonelli, C., 1997. The economics of path-dependence in industrial organization. International Journal of Industrial Organization 15, 643-675.
Antonelli, C., 2007. Technological knowledge as an essential facility. Journal of Evolutionary Economics 17, 451-471.
Antonelli, C., 2008. Localized technological change. Towards the economics of complexity. Routledge, London and New York.
Arocena, R., Sutz, J., 1999. Looking at National Systems of Innovation from the South, DRUID Summer Conference 1999.
Arundel, A., Hollanders, H., 2005. EXIS: An Exploratory Approach to Innovation Scoreboards. European Commission Enterprise Directorate-General.
Arundel, A., Lorenz, E., Lundvall, B.Å., Valeyre, A., 2007. How Europe's economies learn: a comparison of work organization and innovation mode for the EU-15. Industrial and Corporate Change 16, 1175-1210.
Arvanitis, S., Hollenstein, H., 1998. Innovative Activity and Firm Characteristics - A Cluster Analysis with Firm-level Data of Swiss Manufacturing, 25th Annual Conference of the European Association for Research in Industrial Economics Copenhagen, 27-30 August 1998.
Azpiazu, D., Notcheff, H., 1995. El desarrollo ausente. Restricciones al desarrollo, neoconservadurismo y élite económica en la Argentina. Ensayos de Economía Política, FLACSO, Buenos Aires.
Bartus, T., 2005. Estimation of marginal effects using margeff. The Stata Journal 5, 309–329.
Beccaria, L., 2007. El mercado de trabajo luego de la crisis. Avances y desafíos, in: Kosacoff, B. (Ed.), Crisis, recuperación y nuevos dilemas. La economía argentina 2002-2007. CEPAL, Chile.
Bell, M., Albu, M., 1999. Knowledge Systems and Technological Dynamism in Industrial Clusters in Developing Countries. World Development, Vol. 27, No. 9, pp. 1715-34.
Benavente, J.M., 2004. Investigación y Desarrollo, Innovación y Productividad: un análisis econométrico a nivel de la firma, Econometric Society 2004 Latin American Meetings 269.
Benavente, J.M., Contreras, J.L., 2006. Margen de Utilidad y Distribución de las Rentas: La Industria Manufacturera Chilena durante los Noventa. Trimestre Económico, LXXVI (1) : 119-144.
Bibliography
- 251 -
Bernard, A., Jensen, J.B., Redding, S., Schott, P., 2007. Firms in International Trade. Journal of Economic Perspectives 21, 105–130.
Bisang, R., 1993. Industrialización e Incorporación del progreso técnico. Hacia la articulación de un sistema nacional de innovación. CEPAL Buenos Aires, Documento de trabajo Nro 54.
Bisang, R., 1995. Libremercado, intervenciones estatales e instituciones de Ciencia y Técnica. Revista Redes 3.
Bisang, R., Lugones, G., Yoguel, G.e., 2002. Apertura e Innovación en la Argentina. Para desconcertar a Vernon, Schumpeter y Freeman. Miño y Dávila, Buenos Aires.
Blackman, A.W., Seligman, E.J., Sogliero, G.C., 1973. An innovation index based on factor analysis Technological Forecasting and Social Change 4, 301-316.
Borello, J., Erbes, A., Robert, V., Roitter, S., Yoguel, G., 2005. The technical skills of information technology workers in Argentina. CEPAL Review 87, 129-146.
Borello, J., Erbes, A., Yoguel, G., 2006. Sistemas locales de innovación y sistemas productivos locales. Universidad Nacional de General Sarmiento Mimeo.
Bottazzi, G., Dosi, G., Jacoby, N., Secchi, A., Tamagni, F., 2010. Corporate performances and market selection: some comparative evidence. Industrial and Corporate Change 19 (6): 1953.
Breschi, S., Malerba, F., 1997. Sectorial Innovation Systems: Technological Regimes, Schumpeterians Dynamics and Spatial Boundaries, in: Edquist, C. (Ed.), Systems of Innovation: Technologies, Institutions and Organizations. Printer, London.
Brown, M.B., Benedetti, J.K., 1977. Sampling Behavior of Tests for Correlation in Two-Way Contingency Tables. Journal of the American Statistical Association 72, 309-315.
Bunge, A., 1940. Una nueva Argentina. Hyspamérica Ediciones Argentinas SA., Buenos Aires.
Bush, V., 1945. Science The Endless Frontier. A Report to the President. United States Government Printing Office, Washington.
Canitrot, A., 1975. La experiencia populista de redistribución de ingresos. Desarrollo Económico 15, 331-351.
Canitrot, A., 1980. La disciplina como objetivo de la politica económica. Un ensayo sobre el programa económico del gobierno argentino desde 1976. Desarrollo Económico 19, 453-475.
Cassiolato, J., Lastre, H., Szapiro, M., 2000. Arranjos e sistema produtivos locais e proposições de políticas de desenvolvimento industrial e tecnológico, Seminário Local Clusters, Innovation Systems and Sustained Competitiveness. Rio de Janeiro: IE-BNDES, Nota Técnica 5.
Cassiolato, J., Lastre, H., 2002. Systems of innovation and development from a South American perspective: a contribution to Globelics, Globelics WP.
Cassiolato, J., Lastre, H., 2005. Innovation systems and local productive arrangements: New strategies to promote the generation, acquisition and diffusion of knowledge. Innovation: Management, Policy & Practice 7, 172-187.
Cassiolato, J., Lastre, H., 2008. Discussing innovation and development: Converging points between the Latin American school and the Innovation Systems perspective? Globelics WP 08-02.
Castellacci, F., 2010. How does competition affect the relationship between innovation and productivity? Estimation of a CDM model for Norway. Economics of Innovation and New Technology 20, 637-658.
Cefis, E., Orsenigo, L., 2001. The persistence of innovative activities: A cross-countries and cross-sectors comparative analysis. Research Policy 30, 1139-1158.
CEP, 2012. Fichas sectoriales: Estructura y evolución 1991 - 2010, in: Centro de Estudios de la Producción (CEP), M.d.I. (Ed.), Argentina.
CEPAL, 2008. La transformación productiva 20 años después. Viejos problemas, nuevas oportunidades. Naciones Unidas, CEPAL, Santiago de Chile.
Cesaratto, S., Mangano, S., 1993. Technological Profiles And Economic Performance In The Italian Manufacturing Sector. Economics of Innovation and New Technology 2, 237-256.
Bibliography
- 252 -
Cimoli, M., 2005. Heterogeneidad estructural, asimetrías tecnológicas y crecimiento en América Latina. Naciones Unidas, CEPAL, Santiago de Chile.
Clausen, T., Pohjola, M., Sapprasert, K., Verspagen, B., 2011. Innovation strategies as a source of persistent innovation. Industrial and Corporate Change 22, 33-72.
Coase, R., 1937. The Nature of the Firm. Economica 4, 386-405.
Coad, A., Rao, R., 2011. The firm-level employment effects of innovations in high-tech US manufacturing industries. Journal of Evolutionary Economics 21, 255-283.
Cohen, W., Levinthal, D., 1990. Absorptive Capacity: A New Perspective on Learning and Innovation. Administrative Science Quarterly 35, 128-152.
Combs, R., Saviotti, P., Walsh, V., 1987. Economics and Technological Change. Mac Millan, Chicago.
Conover, W.J., 1999. Practical Nonparametric Statistics. 3rd ed. Wiley, New York.
Cortes Conde, R., 1963. Problemas del crecimiento industrial de la Argentina (1870-1914). Desarrollo Económico 12.
Cortes Conde, R., 2005. La economía política de la Argentina en el siglo XX. EDHASA, Buenos Aires.
Crépon, B., Duguet, E., Mairesse, J., 1998. Research, innovation and productivity: an econometric analysis at the firm level. NBER WP 6696.
Crespi, G., Criscuolo, C., Haskel, J., Slaughter, M., 2007. Productivity Growth, Knowledge Flows and Spillovers. Centre for Economic Performance, LSE, Discussion Papers dp0785.
Chamberlain, G., 1980. Analysis of Covariance with Qualitative Data. Review of Economic Studies 47 225-238.
Chamberlain, G., 1984. Panel Data, in: Griliches, Z., Intriligator, M.D. (Eds.), Handbook of Econometrics. North Holland, Amsterdam, pp. 1247-1318.
Chamberlain, G., 1992. Comment: Sequential Moment Restrictions in Panel Data. Journal of Business and Economic Statistics 10, 20-26.
Chandler, A., 1966. Strategy and Structure. Doubleday & Co. Anchor Books Edition, New York.
Chudnovsky, D., López, A., Pupato, G., 2004. Innovation and productivity: A study of Argentine manufacturing frms´behavior (1992-2001). Depto Economía, Universidad de San Andrés, WP 70.
Chudnovsky, D., López, A., Rossi, M., Ubfal, D., 2006. Evaluating A Program of Public Funding of Private Innovation Activities. An Econometric Study of FONTAR in Argentina. Inter-American Development Bank, Office of Evaluation and Oversight (OVE).
Dalum, B., 1992. Export Specialisation, Structural Competitiveness and National Systems of Innovation, in: Lundvall, B.Å. (Ed.), National Systems of Innovation. Towards a Theory of Innovation and Interactive Learning. Pinter Publishers, London.
David, P., 1985. Clio and the Economics of QWERTY. The American Economic Review 75, 332-337.
David, P., 2007. Path dependence: a foundational concept for historical social science. Cliometrica, Journal of Historical Economics and Econometric History 1, 91–114.
Davies, S., 2011. Regional resilience in the 2008–2010 downturn: comparative evidence from European countries. Cambridge Journal of Regions, Economy and Society 4, 369-382.
de Jong, J.P.J., Marsili, O., 2006. The fruit flies of innovations: A taxonomy of innovative small firms. Research Policy 35, 213-229.
De Negri, J.A., Saleno, M.S., Barros de Castro, A., 2005. Innovações, padrões tecnológicos e desempenho das firmas industriais brasileiras, in: De Negri, J.A., Saleno, M.S. (Eds.), Innovações, padrões tecnológicos e desempenho das firmas industriais brasileiras. IPEA, Brazil.
Diamand, M., 1972. La estructura productiva desequilibrada argentina y el tipo de cambio. Desarrollo Económico 12.
Bibliography
- 253 -
Diamand, M., 1983. El péndulo Argentino. UIA; Buenos Aires, Argentina.
Diaz Alejandro, C., 1967. The Argentine Tariff, 1906-1940. Oxford Economic Papers 19, 75-98.
Diaz Alejandro, C., 1975. Ensayos sobre la historia económica argentina. Amorrortu Editores, Buenos Aires.
Dosi, G., 1982. Technological paradigms and technological trajectories: A suggested interpretation of the determinants and directions of technical change. Research Policy 11, 147-162.
Dosi, G., 1988. Sources, Procedures and Microeconomic Effects of Innovation. Journal of Economic Literature 26, 1120-1171.
Dosi, G., 1991. Una reconsideración de las condiciones y los modelos del desarrollo. Una perspectiva 'evolucionista' de la innovación, el comercio y el crecimiento. Pensamiento Iberoamericano, Revista de Economía Política 20.
Dosi, G., Lechevalier, S., Secchi, A., 2010. Introduction: Interfirm heterogeneity—nature, sources and consequences for industrial dynamics. Industrial and Corporate Change 19, 1867-1890.
Drucker, P., 1964. Managing for Results. Harper & Row, New York.
Dutrenit, G., 2009. Sistemas Regionales de Innovación: un espacio para el desarrollo de las PyMEs. El caso de las maquiladoras industriales. UAM-Textual ediciones, Mexico.
Dutrenit, G., Capdevielle, M., Corona Alcantar, J.M., Puchet Anyul, M., Santiago, F., Vera-Cruz, A., 2010. El sistema nacional de innovación mexicano: instituciones, políticas, desempeño y desafíos. Universidad Autónoma Metropolitana. Textual Ediciones, Mexico.
Dutrenit, G., Katz, J., 2005. Innovation, growth and development in Latin-America: Stylized facts and a policy agenda. Innovation: Management, Policy & Practice 7, 105-130.
Dutrenit, G., Puchet, M., 2011. Approaching the measurement of the critical mass of science, technology and innovation: How far off is Mexico?, Globelics 2011, Buenos Aires.
Edquist, C., 2001. Systems of Innovation for Development (SID). Background Paper for Chapter I: “Competitiveness, Innovation and Learning: Analytical Framework” for the UNIDO World Industrial Development Report (WIDR).
Edquist, C., 2004. Systems of innovation: perspectives and challenges, in: Fagerberg, J., Mowery, D., Nelson, R. (Eds.), The Oxford Handbook of Innovation. Oxford University Press, USA.
Edquist, C., Lundvall, B.Å., 1993. Comparing the Danish and Swedish systems of innovation, in: Nelson, R. (Ed.), National Innovation Systems: A Comparative Analysis. Oxford University Press, USA.
Erbes, A., Motta, J., Roitter, S., Yoguel, G., 2004. La construcción de competencias tecnológicas en la fase de crisis del Plan de Convertibilidad, RED PYMES, 9ª Reunión Anual Pymes-Mercosur: "El Rol de las Pequeñas y Medianas Empresas en un nuevo modelo de desarrollo", Buenos Aires.
Erbes, A., Robert, V., Yoguel, G., 2010. Capacities, innovation and feedbacks in production Networks in Argentina. Economics of Innovation and New Technology 19, 719-741.
Eurostat, 2012. Eurostat Statistics Database. In: http://epp.eurostat.ec.europa.eu/.
Fagerberg, J., 2003. Innovation: A Guite to the Literature. Oslo University, Centre for Technology, Innovation and Culture.
Fagerberg, J., Mowery, D., Nelson, R., 2004. The Oxford Handbook of Innovation. Oxford University Press, USA.
Fajnzylber, F., 1989. Industrialización de América Latina: de la 'caja negra' al 'casillero vacío. Cuadernos de la CEPAL 60.
Ferraz, J., Kupfer, D., Haguenauer, L., 1996. Made in Brazil : desafios competitivos para a indu stria. Editora Campus, Rio de Janeiro, Brazil.
Ferrer, A., 2004. La Economía Argentina. Desde sus orígenes hasta principios del siglo XXI Primera Edicion 1963. Fondo de Cultura Economica., Argentina.
Fisher, R., 1936. The use of multiple measurements in taxonomic problems. Annals of Eugenics 7, 179–188.
Bibliography
- 254 -
Fix, E., Hodges, J.L., 1951. Discriminatory analysis: Nonparametric discrimination, consistency properties. . In Technical Report No. 4, Project No. 21-49-004. Randolph Field, Texas: Brooks Air Force Base, USAF School of Aviation Medicine.
Fraga, M., Martins, O., Anciaes, P., 2008. Patterns of innovation diffusion and technological competition in Portuguese manufacturing and service industries. International Review of Applied Economics 22, 353-372.
Freeman, C., 1974. The Economics of Industrial Innovation. Penguin modern economic text, UK.
Freeman, C., 1982a. Technological Infraestructure and International Competitiveness. Draft paper submitted to the OECD ad hoc group on science, technology and competitiveness.
Freeman, C., 1982b. Schumpeter or Schmookler?, in: Freeman, C., Clark, J., Soete, L. (Eds.), Unemployment and Technical Innovation. A study of Long Waves and Economic Development. Pinter, London.
Freeman, C., 1987. Technology policy and economic performance: Lessons from Japan. Pinter, UK.
Freeman, C., 1995. The ‘National System of Innovation’ in historical perspective. Cambridge Journal of Economics 19, 5-24.
Freeman, C., Soete, L., 1997. The economics of industrial innovación. 3rd ed. Pinter, London.
Freeman, C., 2002. Continental, national and sub-national innovation systems—complementarity and economic growth. Research Policy 31, 191-211.
Frenkel, R., Rapetti, M., 2011. Fragilidad externa o desindustrialización: ¿Cuál es la principal amenaza de América Latina en la próxima década? CEPAL, Serie macroeconomía del desarrollo 116.
Frenz, M., Lambert, R., 2009. Exploring Non-technological and Mixed Modes of Innovation Across Countries, in: OECD (Ed.), Innovation in Firms. A Microeconomic Perspective. OECD.
Galeano, E., 1998. Las venas abiertas de América Latina. Siglo XXI Editores. 18th edition, Argentina.
Garud, R., Kumaraswamy, A., Karnøe, P., 2010. Path Dependence or Path Creation? Journal of Management Studies 47, 760-774.
Geroski, P.A., Van Reenen, J., Walters, C.F., 1997. How persistently do firms innovate? Research Policy 26, 33-48.
Godinho, M., Mendonca, S., Pereira, T., 2004. Towards a Taxonomy of Innovation Systems. Globelics 2004, Beijing.
Goedhuys, M., 2007. The impact of innovation activities on productivity and firm growth: evidence from Brazil. ONU-MERIT, WP.
Gutman, G., Ríos, P., 2010. Dinámica reciente de la industria láctea en Argentina. Estrategias de innovación y de integración regional, IDRC - Centro Redes - CEFIR, Buenos Aires-Montevideo.
Gutman, G., Lavarello, P., 2010. Desarrollo reciente de la moderna biotecnología en el sector de la salud humana. Documento de trabajo CEUR, 3/2010. Setiembre.
Hall, B., Lotti, F., Mairesse, J., 2006. Employment, Innovation, and Productivity: Evidence from Italian Microdata. ONU-MERIT, WP.
Hassink, R., 2010. Regional resilience: a promising concept to explain differences in regional economic adaptability? Cambridge Journal of Regions, Economy and Society 3, 45-58.
Herrera, A., 1971. Ciencia y política en América Latina. Siglo XXI, Mexico.
Hervas Oliver, J.L., Sempere Ripoll, F., Boronat Moll, C., 2012. Process innovation objectives and management complementarities: patterns, drivers, co-adoption and performance effects. UNU-MERIT WP 2012-051.
Hirschman, A., 1968. La economía política de la industrialización a través de la sustitución de importaciones. El trimestre económico 35.
Bibliography
- 255 -
Hollenstein, H., 2000. Innovation Modes in the Swiss Service Sector A Cluster Analysis Based on Firm-level Data, 3rd Workshop of the Focus Group on Innovative Firms and Networks. OECD.
Huergo, E., Moreno, L., 2011. Does history matter for the relationship between R&D, innovation, and productivity? Industrial and Corporate Change 20, 1335-1368.
Iacovone, L., Crespi, G., 2010. Catching up with the technological frontier:Micro-level evidence on growth and convergence. Industrial and Corporate Change 19, 2073-2096.
INDEC, 1998. Encuesta sobre la Conducta Tecnológica de las Empresas Industriales Argentinas 1992/1996. INDEC, Buenos Aires. http://www.indec.gov.ar/nuevaweb/cuadros/16/enit_98-01.pdf.
INDEC, 2003. Segunda Encuesta Nacional de Innovación y Conducta Tecnológica de las Empresas (1998-2001). INDEC - SECyT - CEPAL, Buenos Aires.
INDEC, 2006. Encuesta Nacional a Empresas sobre Innovación, I+D y TICs. 2002-2004. SECYT-INDEC, Buenos Aires. http://www.indec.gov.ar/nuevaweb/cuadros/16/enit_02-04.pdf.
INDEC, 2007. Base de datos de desempeño empresarial (BDDE). INDEC-MECON, Buenos Aires.
INDEC, 2008. Encuesta Nacional sobre Innovación y Conducta Tecnológica. ENIT 2005. INDEC, Buenos Aires. http://www.indec.gov.ar/nuevaweb/cuadros/16/publicacion-enit-2005.pdf.
INDEC, 2010. Base Integrada de Desempeño Empresarial. INDEC, Buenos Aires.
Ito, K., Lechevalier, S., 2010. Why some firms persistently out-perform others: investigating the interactions between innovation and exporting strategies. Industrial and Corporate Change 19, 1997-2039.
Jensen, M.B., Johnson, B., Lorenz, E., Lundvall, B.Å., 2007. Forms of knowledge and modes of innovation. Research Policy 36, 680-693.
Johnson, B., 1992. Institutional Learning, in: Lundvall, B.Å. (Ed.), National Systems of Innovation - Towards a Theory of Innovation and Interactive Learning. Pinter, London.
Johnson, B., Edquist, C., Lundvall, B.Å., 2003. Economic Development and the National System of Innovation Approach, Globelics 2003, Rio de Janeiro.
Jonckheere, A.R., 1954. A distribution free k-sample test against ordered alternatives. Biometrika 41, 133-145.
Juneau, P., 2006. Nonparametric Inference for Ordered Alternatives in a One-Way Layout Using the SAS System. Paper SD11. Pfizer Global Research & Development, Ann Arbor, Michigan.
Katz, J., 1976. Creación de Tecnología en el sector manufacturero argentino. BID/CEPAL. Buenos Aires.
Katz, J., 2000. Pasado y presente del comportamiento tecnológico de América Latina. CEPAL Santiago de Chile, Serie Desarrollo Económico 75.
Katz, J., 2007. Reformas estructurales orientadas al mercado, la globalización y la transformación de los sistemas de innovación en América Latina, in: Dutrenit, G., Jasso, J., Villavicencio, D. (Eds.), Globalización, acumulación de capacidades e innovación. Los desafíos para las empresas, localidades y países. FCE-OEI, Mexico.
Katz, J., Bernat, G., 2011. Macroeconomic adjustment and structural change during the last decade. Argentina, Brazil and Chile, Globelics 2011, Buenos Aires.
Katz, J., Kosacoff, B., 1998. Aprendizaje tecnológico, desarrollo institucional y la microeconomía de la sustitución de importaciones. Desarrollo Económico 37, 483-502.
Katz, J., Stumpo, G., 2001. Regímenes sectoriales, productividad y competitividad internacional. Revista de la CEPAL 75, Santiago de Chile.
Kaufman, L., Rousseeuw, P.J., 1990. Finding Groups in Data: An Introduction to Cluster Analysis. . Wiley, New York.
Kemp, R.G.M., De Jongm, P.J., Folkeringa, M., Wubben, E.F.M., 2003. Innovation and firm performance. Differences between small and medium-sized firms. SCALES-paper N200213.
Bibliography
- 256 -
Kendall, M.G., 1955. Rank Correlation Methods. . Hafner Publishing Co, New York.
Keynes, J.M., 1936. The General Theory of Employment, Interest and Money. Macmillan. London. (Spanish version 1974, Fondo de Cultura Económica, Mexico).
Klecka, W.R., 1980. Discriminant Analysis. Newbury Park, USA.
Kline, S., Rosenberg, N., 1989. An Overview of Innovation, in: Landau, R., Rosenberg, N. (Eds.), The Positive Sum Strategy: Harnessing Technology for Economic Growth. National Academies Press, Washington.
Kosacoff, B., 1996. Estrategias empresariales en la transformación industrial argentina. CEPAL Buenos Aires, Documento de Trabajo 67.
Kosacoff, B., 1998. Estrategias empresariales en tiempos de cambio, in: Kosacoff (Ed.), Estrategias empresariales en tiempos de cambio. CEPAL, Universidad Nacional de Quilmes, Buenos Aires.
Kosacoff, B., Heyman, D., 2000. La Argetina de los Noventa. EUDEBA-CEPAL, Buenos Aires.
Kosacoff, B., Ramos, D., 2006. Comportamientos microeconómicos en entornos de alta incertidumbre: la industria argentina. CEPAL Buenos Aires, Documento de Proyecto.
Kruskal, W.H., Wallis, W.A., 1952. Use of Ranks in One-Criterion Variance Analysis. Journal of the American Statistical Association 47, 583-621.
Lachenbruch, P.A., 1975. Discriminant analysis. Hafner, New York.
Lall, K., Dunnewijk, T., 2008. Entrepreneurship and Innovation Strategies in ICT SMEs in Enlarged Europe (EU25). United Nations University - Maastricht Economic and social Research and training centre on Innovation and Technology, WP.
Lall, S., 1984. Technology exports of newly industrializing countries. World Development Report 12.
Lall, S., 1992. Technological capabilities and industrialization. World Development Report 20.
Lall, S., 2001. Competitiveness, Technology and Skills. Edward Elgar, Chettenham.
Langlois, R.N., 2002. Schumpeter and the Obsolescence of the Entrepreneur. SSRN eLibrary.
Langlois, R.N., Robertson, P.L., 1995. Firms, Markets and Economic Change: A dynamic Theory of Business Institutions. Routledge, London and New York.
Lanjouw, J.O., Schankerman, M., 2004. Patent Quality and Research Productivity: Measuring Innovation with Multiple Indicators*. The Economic Journal 114, 441-465.
Le Bas, C., Mothe, C., Nguyen Thi, T.U., 2011. Technological Innovation Persistence: Literature Survey and Exploration of the Role of Organizational Innovation. SSRN eLibrary.
Leiponen, A., Drejer, I., 2007. What exactly are technological regimes?: Intra-industry heterogeneity in the organization of innovation activities. Research Policy 36, 1221-1238.
List, F., 1841. Das Nationale System der Politischen Ökonomie. Basel: Kyklos (translated and published under the title: The National System of Political Economy' by Longmans, Green and Co, London.
López, A., 2003. Industrialización sustitutiva de importaciones y sistema nacional de innovación: un análisis del caso argentino. Redes. Revista de Estudios Sociales de la Ciencia 19, Buenos Aires.
López, A., Arza, V., 2008. Characteristics of university-industry linkages in the Argentinean industrial sector. Globelics México 2008, México D.F.
Lovio, R., Mickwitz, P., Heiskanen, E., 2011. Path Dependence, Path Creation and Creative Destruction in the Evolution of Energy Systems in: Wüstenhagen, R., Wuebker, R. (Eds.), Handbook of Research on Energy Entrepreneurship. Edward Elgar Publishing, Inc, UK.
Lugones, G., Suárez, D., 2010. STI indicators for policy making in developing countries: An overview of experiences and lessons learned, Conference Room Paper, Multi-Year Expert Meeting on Enterprise development policies and capacity-building in ST&I, Ginebra.
Lugones, G., Suárez, D., 2006. Los magros resultados de las políticas para el cambio estructural en América Latina: problema instrumental o confusión de objetivos? Centro Redes, WP 27.
Bibliography
- 257 -
Lugones, G., Suárez, D., Le Clech, N., 2007. Innovative Behaviour and its impact on firms´ performance, Micro Evidence on innovation in developing countries, UNU-MERIT, Maastricht.
Lugones, G., Suárez, D., Moldován, P., 2008. Innovation, competitiveness and salaries: a model of combined growth at the firm level. Globelics México 2008, México D.F.
Lundvall, B.Å., 1992. National System of Innovation: Towards a Theory of Innovation and Interactive Learning. Pinter, London.
Lundvall, B.Å., 1996. The social dimension of the learning economy. DRUID WP 96-1.
Lundvall, B.Å., 1998. Defining Industrial Dynamics and its Research Agenda, DRUID winter seminar, Middelfart.
Lundvall, B.Å., 2004. Why the New Economy is a Learning Economy? DRUID WP 04-01.
Lundvall, B.Å., Lam, A., 2007. The Learning organisation and national systems of competence building and innovation, in: Lorenz, N., Lundvall, B.Å. (Eds.), How Europe's Economies Learn: Coordinating Competing Models. Oxford University Press, USA.
Lundvall, B.Å., 2009. Investigación en el campo de los sistemas de innovación: orígenes y posible futuro, in: Lundvall, B.Å. (Ed.), Sistemas Nacionales de innovación. Hacia una teoría del aprendizaje por interacción. UNSAM, Argentina.
Lundvall, B.Å., Joseph, K.J., Chaminade, C., Vang, J., 2009a. Handbook On Innovation Systems And Developing Countries. Edward Elgar.
Lundvall, B.Å., Chaminade, C., Vang-Lauridsen, J., Joseph, K.J., 2009b. Innovation policies for development: towards a systemic experimentation based approach, Globelics 2009, Dakar.
Lynch, R., 2009. Strategic Management. Pearson Longman.
Mahalanobis, P.C., 1936. On the generalized distance in statistics. National Institute of Science of India 12: 49–55.
Mairesse, J., Sassenou, M., 1991. R&D productivity: a survey of econometric studies at the firm level. National Bureau of Economic Research, NBER WP 3666.
Malerba, F., Nelson, R., 2007. Catching up in different sectoral systems, Globelics 2007, Saratov.
Malerba, F., 2004. Sectoral Systems: how and why innovation differs across sectors, in: Fagerberg, J., Mowery, D., Nelson, R. (Eds.), The Oxford Handbook of Innovation. Oxford University Press, USA.
Malerba, F., Orsenigo, L., 1997. Technological Regimes and Sectoral Patterns of Innovative Activities. Industrial and Corporate Change 6, 83 – 118.
Malerba, F., Orsenigo, L., Peretto, P., 1997. Persistence of innovative activities, sectoral patterns of innovation and international technological specialization. International Journal of Industrial Organization 15, 801-826.
Mann, H.B., Whitney, D.R., 1947. On a Test of Whether one of Two Random Variables is Stochastically Larger than the Other. The Annals of Mathematical Statistics 18, 50-60.
Mansfield, E., 1962. Entry, Gibrat's Law, Innovation, and the Growth of Firms. The American Economic Review 52, 1023-1051.
Martin, R., 2011. Regional economic resilience, hysteresis and recessionary shocks. Journal of Economic Geography, 1–32.
Marx, K., 1858. El Capital. Crítica de la economía política. Siglo XXI Editores. México, 1975.
Meade, J.E., 1952. External Economies and Diseconomies in a Competitive Situation. The Economic Journal 62, 54-67.
Mensch, G., 1975. Stalemate in Technology: Innovations Overcome the Depression. Ballinger, Cambridge.
Metcalfe, J.S., 1995. Technology systems and technology policy in an evolutionary framework Cambridge Journal of Economics 19, 25-46.
Bibliography
- 258 -
Metcalfe, J.S., 2002. Restless capitalism increasing returns and growth in evolutionary enterprise economics, Organisational and technological transformations in complex adaptive systems: concepts, methods of analysis and measurement, Manchester, UK.
Milesi, D., 2006. Patrones de innovación en la industria manufacturera argentina: 1998-2001, Universidad Nacional de General Sarmiento, Documento de Trabajo LITTEC.
Mintzberg, 1983. Structure in 5's: Designing Effective Organizations. Prentice Hall. USA.
Mintzberg, H., Waters, J., 1985. Of strategies, deliberate and emergent. Strategic Management Journal 6, 257-272.
Moran, B., Tame, P., 2012. Organizational Resilience: Uniting Leadership and Enhancing Sustainability. Sustainability: The Journal of Record 5, 233-237.
Mowery, D., Rosenberg, N., 1982. The influence of market demand upon innovation: a critical review of some recent empirical studies, in: Rosenberg, N. (Ed.), Inside the Black Box: Technology and Economics. Cambridge University Press.
Myrdal, G., 1974. What Is Development? Journal of Economic Issues 8, 729-736.
Narula, R., 2003. Understanding Absorptive Capacities in an “Innovation Systems” Context: Consequences for Economic and Employment Growth. DRUID WP 04-02.
Nelson, R., 1991a. Why do firms differ, and how does it matter? Strategic Management Journal 12, 61-74.
Nelson, R., 1991b. The role of firm differences in an evolutionary theory of technical advance. Science and Public Policy 18, 347-352.
Nelson, R., 1993. National Innovation Systems: A Comparative Analysis. Oxford University Press, USA.
Nelson, R., 1994. The Co-evolution of Technology, Industrial Structure, and Supporting Institutions. Industrial and Corporate Change 3, 47-63.
Nelson, R., 1995. Recent evolutionary theorizing about economic change. Journal of Economic Literature 33, 48-90.
Nelson, R., Dahlman, C., 1991. Social absorption capability, national innovation systems, and economic development, paper prepared for the Korean Development Institute´s Symposium. Kore, July 1-2.
Nelson, R., Winter, S., 1977. In search of useful theory of innovation. Research Policy 6, 36-76.
Nelson, R., Winter, S., 1982. An evolutionary theory of economic change. The Belknap Press of Harvard University Press, Cambridge.
Novick, M., Yoguel, G., 2001. La vulnerabilidad de una trama productiva: la díficil relación cliente-proveedor en el complejo automotriz argentino. Revista Trabajo 2-4.
Nurkse, R., 1961. Equilibrium and Growth in the World Economy. Harvard University Press.
Ocampo, J.A., 2001. Retomar la agenda del desarrollo. Revista de la CEPAL 74.
Ocampo, J.A., 2012. La historia y los retos del desarrollo latinoamericano. CEPAL, Santiago de Chile.
OECD, 1997. Revision of the High-Technology Sector and Product Classification. OECD.
OECD, 2005. Oslo Manual - 3rd edition. Guidelines for collecting and interpreting innovation data. First edition 1992. OECD.
OECD, 2007. Science, Technology and Innovation Indicators in a Changing World. Responding to Policy Needs. OECD. Brussels.
Østergaard, C., Holm, J., 2010. Sources of regional resilience in the Danish Sector, Druid Summer Conference 2010. "Opening Up Innovation: Strategy, Organization and Technology" Imperial College London Business School, June.
Østergaard, C., Park, E.K., 2013. Cluster Decline and Resilience - The Case of the Wireless Communication Cluster in North Jutland, Denmark. Available at http://ssrn.com.
Bibliography
- 259 -
Panos, S., 2008. State dependence in work-related training participation among British employees: A comparison of different random effects probit estimators. MPRA Paper 14261, University Library of Munich, Germany, revised Mar 2009.
Pavitt, K., 1984. Sectoral patterns of technical change: towards a taxonomy and a theory. Research Policy 13, 353-369.
Pearson, K., 1900. On the criterion that a given system of deviations from the probable in the case of a correlated system of variables is such that it can be reasonably supposed to have arisen from random sampling. . Philosophical Magazine 5, 157-175.
Penrose, E., 1959. The theory of the growth of the firm. Oxford University Press, USA.
Percival, J.C., Cozzarin, B.P., 2008. Complementarities Affecting the Returns to Innovation. Industry and Innovation 15, 371-392.
Perez, C., Soete, L., 1988. Catching Up in Technology: entry Barriers and Windows of Opportunity, in: Dosi, G., Freeman, C., Nelson, R., Soete, L. (Eds.), Technical Change and Economic Theory. Francis Pinter, London.
Peters, B., 2009. Persistence of innovation: stylised facts and panel data evidence. The Journal of Technology Transfer 34, 226-243.
Phillips, A., 1971. Technology and Market Structure: A Study of the Aircraft Industry. Heath, Lexington, Mass.
Molina-Domene, M.A., Pietrobelli, C., 2012. Drivers of technological capabilities in developing countries: An econometric analysis of Argentina, Brazil and Chile. Structural Change and Economic Dynamics 23, 504-515.
Pillai, K.C.S., 1985. Multivariate analysis of variance (MANOVA), in: Kotz, S., Johnson, N.L., Read, C.B. (Eds.), Encyclopedia of Statistical Sciences (vol 6). Wiley, New Youk.
Porta, F., 2006. Especialización productiva e inserción internacional. Evidencias y reflexiones sobre el caso argentino, in: Lugones, G., Porta, F. (Eds.), Enfoques y metodologías alternativas para la medición de las capacidades innovativas. FONCYT-ANPCYT, Buenos Aires.
Porta, F., Bonvecchi, C., 2003. Las condiciones de consistencia micro/macroeconómica. REDES-CECE, Préstamo BID 925/OC-AR. Pre II. CEPAL Buenos Aires.
Porta, F., Suárez, D., De Angelis, J., 2011. Sistemas nacionales de innovación en el MERCOSUR: convergencias y asimetrías, in: Caetano, G. (Ed.), MERCOSUR. 20 años. CEFIR, Montevideo.
Porter, M., 1985. Competitive Advantage: Creating and Sustaining Superior Performance. Free Press NY.
Porter, M., 1990. La Ventaja Competitiva de las Naciones. Vergara.
Prebisch, R., 1950. El desarrollo económico de la América Latina y algunos de sus principales problemas Desarrollo Económico 26.
Rahmeyer, F., 2007. From a Routine-based to a Knowledge-based View: Towards an Evolutionary Theory of the Firm, in: Hanusch, H., Pyka, A. (Eds.), Elgar Companion To Neo-Schumpeterian Economics. Edward Elgar Publishing.
Raymond, W., Mohnen, P., Palm, F., Schim van der Loeff, S., 2007. The Behavior of the Maximum Likelihood Estimator of Dynamic Panel Data Sample Selection Models CESifo WP 1992.
Raymond, W., Mohnen, P., Palm, F., van der Loeff, S.S., 2010. Persistence of Innovation in Dutch Manufacturing: Is It Spurious? The Review of Economics and Statistics 92, 495-504.
Reinert, E., 1996. The role of technology in the creation of rich and poor nations: underdevelopment in a Schumpeterian system, in: Aldcroft, Catterall (Eds.), Rich nations-poor nations. Elgar, UK.
Reinert, E., Reinert, S., 2003. An Early National Innovation System: The Case of Antonio Serra’s 1613 Breve Trattato Institutions and Economic Development/Istituzioni e Sviluppo Economico 1.
Rencher, A.C., 2002. Methods of Multivariate Analysis. 2nd ed. Wiley, New York.
RICyT, 2000. Manual de Bogotá: Normalización de Indicadores de Innovación Tecnológica en América Latina y el Caribe. OEA/ RICYT/COLCIENCIAS/OCT, Bogotá, Colombia.
Bibliography
- 260 -
RICyT, 2012. Indicadores de Ciencia y Tecnología, in: Tecnología, R.I.d.C.y. (Ed.), www.ricyt.org.
Rivera Ríos, M.A., 2009. Desarrollo económico y cambio institucional. Una aproximación al estudio del atraso económico y el desarrollo tardío desde perspectiva sistémica. UNAM-Casa Editorial Juan Pablos, México.
Robinson, J., 1962. Essays in the Theory of Economic Growth. Macmillan London.
Roitter, S., Erbes, A., Kababe, Y., 2011. Learning processes in services: new forms of work organization? , Globelics 2011, Buenos Aires.
Romer, P.M., 1990. Endogenous Technological Change. Journal of Political Economy 98, 71-102.
Rossegger, G., 1987. The economics of production and innovation. An Industrial Perspective. Pergamon Press.
Rostow, W., 1962. The Stages of Economic Growth. Cambridge University Press, London.
Sábato, J., Botana, N., 1970. La ciencia y la tecnología en el desarrollo de América Latina, Tiempo Latinoamericano. Editorial Universitaria, Santiago de Chile.
Salter, W.G.E., 1960. Productivity and Technical Change. Cambridge University Press, New York.
Schmookler, J., 1979. Fuentes Económicas de la Actividad Inventiva, in: Rosenberg, N. (Ed.), Economía del Cambio Tecnológico. Fondo de Cultura Económica, México.
Schumpeter, J., 1912. The theory of economic development. Harvard University Press, Cambridge.
Schumpeter, J., 1942. Capitalism, socialism, and democracy. Harper and Brothers, New York.
Schumpeter, J., 1947. The creative response in economic history. The Journal of economic history 7, 149-159.
Sen, A., 1999. Development as Freedom. Oxford University Press, USA.
Singer, H., 1950. Distribución de ganancias entre países inversores y prestatarios. American Economic Review, Papers and Proceedings 2.
Smith, A., 1776. Investigación sobre la Naturaleza y Causa de la Riqueza de las Naciones. Ed. Hyspamérica, España, 1983.
Solow, R.M., 1956. A Contribution to the Theory of Economic Growth. The Quarterly Journal of Economics 70, 65-94.
SPU, 2010. Anuario de Estadísticas Universitarias. Ministerio de Educación, Argentina.
Srholec, M., Verspagen, B., 2012. The Voyage of the Beagle into innovation: explorations on heterogeneity, selection, and sectors. Industrial and Corporate Change 21, 1221-1253.
Stubrin, L., 2011. R&D collaborations and innovation performance. The case of Argentinean biotech firms, Globelics 2011, Buenos Aires.
Suárez, D., 2006. National Specificities and Innovation Indicators, Globelics 2006, Kerala.
Suárez, D., 2007. Dinámica innovativa y estructura de vinculaciones en la industria manufacturera argentina, Ponencia presentada en ALTEC 2007.
Suárez, D., 2009. Innovative strategies and their impact on the national innovation system dynamics, Globelics 2009, Dakar.
Sztulwark, S., 2005. Especialización Productiva y Subdesarrollo en el Paradigma Informacional. Una Aproximación a Partir del Caso de las Semillas Transgénicas y su Difusión en la Argentina, XI Seminario Ibero Latinoamericano de Gestión Tecnológica. ALTEC – 2005, Brasil.
Teece, D., Augier, M., 2007. Competencies, Capabilities and the Neo-Schumpeterian Tradition, in: Hanusch, H., Pyka, A. (Eds.), Elgar Companion To Neo-Schumpeterian Economics. Edward Elgar Publishing.
Teece, D., Pisano, G., 1994. The Dynamic Capabilities of Firms: an Introduction. Industrial and Corporate Change 3, 537-556.
Tether, B., 2000. Who Co-Operates For Innovation Within The Supply-Chain, And Why? CRIC Discussion Paper No. 35.
Tether, B., Swann, P., 2003. Sourcing Science. The use by industry of the Science Base for Innovation; Evidence from the UK's Innovation Survey. CRIC Discussion Paper Nº 64.
Bibliography
- 261 -
Tu, C., Han, C., 1982. Discriminant Analysis Based on Binary and Continuous Variables. Journal of the American Statistical Association 77, 447-454.
UIS-UNESCO, 2012. UIS Data Base, in: http://stats.uis.unesco.org/.
Utterback, J., Abernathy, W., 1975. A dynamic model of process and product innovation. Omega, The International Journal of Management Science, Vol.3, Nº6. Pergamon Press, UK.
Villanueva, E., 1972. El origen de la industrialización en la Argentina. Desarrollo Económico Nro 47.
WB, 2012. World Bank Database, in:, http://microdata.worldbank.org.
Wernerfelt, B., 1984. The resource-based view of the firm Strategic Management Journal, 5(2), 171-180.
Williamson, O.E., 1981. The Economics of Organization: The Transaction Cost Approach. American Journal of Sociology 87, 548-577.
Wooldridge, J.M., 2005. Simple solutions to the initial conditions problem in dynamic, nonlinear panel data models with unobserved heterogeneity. Journal of Applied Econometrics 20, 39-54.
Yoguel, G., 2000. Aprendizaje y competencias como factores competitivos en el nuevo escenario: algunas reflexiones desde la perspectiva de la empresa, in: Poma, L., Boscherini, F. (Eds.), Territorio, conocimiento y competitividad de las empresas. Miño y Davila. Buenos Aires.
Yoguel, G., Barletta, F., 2010. La actualidad del pensamiento de Schumpeter. Revista de Trabajo Nueva Época. Teorías económicas y políticas públicas frente a la crisis global 5 - 7.
Yoguel, G., Robert, V., 2010. La dinámica compleja de la innovación y el desarrollo económico. Desarrollo Económico 50, 423-453.
Yoguel, G., Robert, V., Pereira, M., Barletta, F., 2011. The effects of feedbacks on firms’ productivity growth. Micro, macro, and meso determinants of productivity growth in Argentinian firms, Globelics 2011, Buenos Aires, Argentina.
Yoguel, G., Suárez, D., 2011. Hechos estilizados de la vinculación Universidad-Empresa. Revista Apungs, 2011.
Yurtseven, A.E., Tandoğan, V.S., 2012. Patterns of innovation and intra-industry heterogeneity in Turkey. International Review of Applied Economics 26, 657-671.
- 262 -
Appendix
Appendix 1: The panel and the definition of the variables
1.1. Methodology and samples of the National Innovation Surveys
Period of reference
Institution Year Type of
Questionnaire Sample Panel Sample selection Unit of analysis
1998-2001
INDEC
2003
OM – BM
2229 firm
1688 firms Stratification and forced
inclusion.
Manufacturing firms, more than
10 employees
2002-2004 2005 2133 firms
1627 firms
Sample 1998-2001 plus random selection of firms
from the Industrial Register to replace drops.
2005 2007 2167 firms
1675 firms
Sample 2002-2004 plus random selection of firms
from the Industrial Register to replace drops.
2006 (unpublished)
2007 App. 2000 firms
App. 1700 firms
Sample 2005 plus random selection of firms from
the Industrial Register to replace drops.
INDEC: National Institute of Statistics and Census. OM: Oslo Manual, OECD (2005); BM: Bogotá Manual, RICyT (2000). Source: INDEC (2003; 2006; 2007; 2010).
1.2: Definition and matching of the variables100
1.2.1: Innovation expenditure: questions included in each survey
ENIT 1998-2001 ENIT 2002-2004 ENIT 2005-2006
Internal R&D X X X
External R&D X X X
Capital goods X X X
Hardware X X X
Software X X X
Technology transfer X X X
Training X X X
Engineering and industrial design X X X
Consulting X X X
Definitions used in the four surveys (own translation from Spanish):
“Internal Research and development (R&D): creative work performed systematically, that
is, non-occasional, with the objective of generating new knowledge (scientific or technical)
or applying or exploiting already existent one, or developed by others. Three categories of
R&D can be identified: basic research (…); applied research (…) or experimental
development (…). (…) R&D activities are not always performed within an R&D formal
100 Based on INDEC 2003, 2006, 2008 and 2010.
Appendix
- 263 -
laboratory or department. Although it is not an easy task, it is necessary to identify those
R&D activities that are performed without a formal structure. (…).
External Research and Development: creative work (…) commissioned to a third party, by
means of hiring or funding of a group of researchers, institutions or firms with the
agreement of the hiring firm to have total property of results.
Acquisition of capital goods, hardware and/or software: incorporation of goods linked to
the introduction of process, products or organizational or commercial improvements
and/or innovations should be considered innovative activities. The replacement of a
machine or a new version of already installed software does not imply an innovative
activity.
Technology transfer: acquisition of patent, non-patented invents, licenses, trademarks,
designs or know-how rights or the hiring of technical assistance linked to the introduction
of product, process, organizational or commercial improvements or innovations.
Training: it will be considered an innovative activity as long as it does not mean training
new employees in methods, processes or techniques already existent in the firm. Training
can be internal or external, both in soft (management and administration) and hard
(productive processes) technologies.
Engineering and industrial design (EID): technical preparations for production and
distribution, not included in R&D, as well as the blueprints and graphs aimed at defining
procedures, technical specifications and operative characteristics, the installation of
machinery, the performance of industrial engineering and production set up (…).
Consulting: scientific and technical services hiring related to the EID activities or the
development or implementation of informatics systems produced by third parties, externals
to the firms. Services related to R&D or Training should be considered as part of those
activities.
1.2.2. Employment by formal education
ENIT 1998-2001
ENIT 2002-2004
ENIT 2005-2006
a. Employees with basic education or less X X X
b. Employees with technical education X X X
c. Professional employees X X X
d. Total employment (a + b + c) X X X
Appendix
- 264 -
Definitions used in the four surveys (own translation from Spanish):
“Professional employees are those employees who have completed a university.
Employees with technical education are those who have completed the secondary and/or
any tertiary education (not university) and/or uncompleted a university degree.
Employees with basic education or less are those employees who have not completed the
secondary school or less. “
1.2.3. Linkages: Organizations from the NIS
Organizations ENIT 1998-2001 ENIT 2002-
2004 ENIT 2005-2006
Universities and training centers
University + Institute of Technical Training
NC Universities / training centers
Public organizations Agencies and public programs of S&T
promotion NC
ANPCyT + Other governmental organizations for the promotion of
S&T
Headquarters and related firms
Headquarters and firms from the same holding
NC Headquarters / related firms
Unrelated firms Other firms NC Unrelated firms
Suppliers Suppliers NC Suppliers
Clients Clients NC Clients
R&D Labs and enterprises R&D Labs/enterprises NC R&D Labs/enterprises
Technological centers and consultants
Technological center + consultants
NC Technological center/consultants +
INTI + INTA NC: question non-comparable. “+” means that options were merged, “/” means that the options were asked together. ANPCyT: National Agency of Science and Technology Promotion; INTI: National Institute of Industrial Technology; INTA: National Institute of Agricultural Technology.
Definition used in the 1998-2001 and 2006 surveys (own translation from Spanish):
Respondents are asked to indicate “if the firm has had linkages in the context of the
innovative activities (table 1.2.1.)” with a set of predefined organizations, where linkages
have to be understood as “any kind of linkages, either formal (with a contract) or informal
(without a contract) one”, not “restricted to the national scope”.
1.2.4. Access to external resources: type of external organizations
Organizations ENIT 1998-2001 ENIT 2002-2004 ENIT 2005-2006
Public and private banks Public and private banks Banks Banks
Suppliers Suppliers Suppliers Suppliers
Clients Clients Clients Clients
Other firms Other firms Other firms Other firms
Public organisms of innovation promotion
Public organisms of promotion
ANPCyT + other public organizations
ANPCyT + other public organizations
International organizations International organizations
International organizations
International organizations
Others Foundations/ NGO +
universities NGO + Universities +
others NGO + Universities +
others NC: question non-comparable with the previous or following exercise. “+” means that options were merged, “/” means that the options were asked together.
Appendix
- 265 -
Definitions used in the four surveys (own translation from Spanish):
Respondents are ask about the “origin of the funds allocated to support the innovative
activities”, in terms of the percentage of resources accessed from each organization. A row
that includes the level own resources is added so the adding of the percentages is equal to
100%.
1.3: The innovation balance index
In order to analyze impact of different weights in the final innovation balance index (IB),
this appendix presents different estimations. The estimations were run for the whole
dataset without distinguishing the year and categories were analyzed in respect of its
participation in the total expenditures on innovation (the sum of the categories´
participation equals the unity). This means that the estimations were perform over 2400
observations, which reduces the bias of the panel. The categories of innovation expenditure
are the same as the ones presented in Chapter 5: a. Research and development (internal and
external); b. Engineering and industrial design and training; c. Capital goods and hardware;
and d. Technology transfer, consulting and software. In Table 1.3.1, some descriptive
statistics are provided using the weights that arise from an extraction based on factor
analysis, principal component method (FA), from the simple average participation (AP) of
each category in the panel and from the theoretical weights used in chapter 5 (TW).
1.3.1: IB and categories – descriptive statistics (different extraction methods)
Categories (weights) TW FA AP
A 0.250 0.234 0.201
B 0.250 0.224 0.089
C 0.250 0.385 0.553
D 0.250 0.157 0.158
IB
Mean 0.22612488 0.241321167 0.24817676
Median 0.10 1.39 0.178
Standard deviation 0.24393929 0.25298178 0.25594907
Minimum 0 0 0
Maximum 0.9260258 0.980850384 0.9431441
Skwness 0.798 0.715 0.642
Kurtosis -0.521 -0.675 -0.825 Obs: 2400. Source: own elaboration based on INDEC (2010).
The differences between the different extraction methods are relative high in terms of the
weights, especially for the case of category c (capital goods), which is also the category that
explains the bulk of the expenditure on innovation. However, differences are lower in
terms of the resulting index. Might be the only difference worth to mention is the fact that
Appendix
- 266 -
the TW is more restrictive in terms of the general level of the index, so it its more
restrictive in terms of what is a more balanced firm. In this respect, higher levels of the
index will account for firms with more distributed expenditures on innovation, where the
endogenous sources of knowledge (categories a and b) explain a high percentage of the
total expenditure.
Graph 1.3.1 shows the frequencies for each estimations. The peaks of each curve are
explained by those firms which only invested in capital goods or hardware (category c).
Once again, the selected set of weights leads to a more restricted innovation balance.
Graph 1.3.1: Frequency of the IB – different weights (TW, FA and PA)
Obs: 2400. Source: own elaboration based on INDEC (2010).
1.4: Capital ownership – comparison of selected variables
Mean (std. dv) One-way ANOVA (a vs. b)
1% to 99% foreign capitals (a)
100% Foreign Capitals (b)
KO (a+b) F Sig
Sales
98-01 86.0 (198.2) 78.6 (141.5) 82.9 (175.9) 0.390 0.534
02-04 101.4 (281.0) 95.2 (189.2) 98.8 (245.6) 0.294 0.589
05-06 125.9 (345.1) 133.7 (299.0) 129.2 (325.1) 0.073 0.788
Employment
98-01 342 (502) 309 (401) 328 (461) 0.780 0.379
02-04 303 (453) 299 (402) 301 (430) 0.397 0.530
05-06 323 (445) 341 (498) 330 (466) 0.167 0.684
Exports
98-01 14.3 (40.2) 16.1 (37.0) 15.0 (38.8) 1.364 0.245
02-04 24.8 (67.5) 21.5 (51.1) 23.4 (61.0) 1.011 0.317
05-06 67.5 (203.4) 78.3 (285.8) 71.9 (239.8) 0.833 0.363
Expenditure on
innovation
98-01 1.52 (4.19) 1.86 (3.52) 1.66 (3.91) 0.084 0.772
02-04 2.82 (7.26) 2.57 (4.34) 2.72 (6.19) 0.241 0.624
05-06 1.60 (4.27) 1.49 (2.16) 1.55 (3.54) 0.330 0.567
Number of firms 71 51 122 Obs: 800. Sales, exports and expenditures on innovation in millions of constant AR$ (base 1998). Employment in thousands of persons. Source: own elaboration based on INDEC (2010).
0
50
100
150
200
250
300
350
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
FA
TW
PA
Appendix
- 267 -
Appendix 2: The innovative strategies
2.1. Non-parametric and multivariate statistics – Methodology
2.1.1. Krukal-Wallis (KW) statistic
The KS procedure orders the sample based on each variable’s level and the firm gets a
number from 1 up to , being the total number of cases ( =800). Over the ranking, the
KS is contrasted against the null hypothesis of non-significant mean’s differences among
groups. Formally, the KS is estimated as:
C
s
sks
kN
Nn
N
NKW
R
12
2__
12/)1(
)1(2
1
1 .
Where kKW is the value for each variable (3 in this case), sn is the number of
observations within the group (the eight possible combination of innovative strategies),
snN is the number of observations in the sample (800 cases); skR
__
is the average
position in the ranking of the group for the variable .
Since this test looks for at least one different media in eight categories, it is sensitive to
extreme values. Therefore, besides running the test for the whole panel, specific contrast
hypothesis will be added and tested in order to check reduced number of groups.
2.1.2. Jonckheere-Terpstra (JT) statistic
The JT statistic is a generalization of the U statistic of Whitney (1947) for two samples and
it is calculated by the estimation of differences between pairs of contingency tables, under
the null hypothesis of non-significant differences between ordered pairs. The general
notation of JT is:
1
1 2
l
s
I
l
lksk UJT with 2
)1( ss
ksks
nnRU .
Where ks is the value of the variable k for the group s (8 groups), which is also de the
ordering variable, ksR is the sum of the rank of the group s for the variable k , sn is the
number of observations for the group s . The combination of strategies was ordered from
1 to 8, based on the assumption about the cumulative nature of the innovation process,
being past behavior key elements to explain present ones and more continuous strategies
more complex than less continuous strategies. Therefore, strategies were ordered from the
worst (NI) to the best strategy (C) and from past (t-2/t-1) to present activities (t): 0- NI-
NI; 1- NI-N; 2- SP-NI; 3- SP-N; 4- N-SP; 5- N-C; 6- C-SP; 7- C-C. However, since to
establish an a priori ordering of some combinations is not possible (e.g. N-SP versus SP-
Appendix
- 268 -
N), specific contrast hypothesis will be set in order to test what is directly observed
through the rankings.
2.1.3. Pearson’s Chi-square and Kendall´s Tau-c statistics
Pearson’s Chi-square ( will be used to test the null hypothesis of a lack of relationship
between the grouping and the probability of a positive outcome (Link / RE=1) The
statistic can be formalized as: ∑
.
Where is sum of cell weights for cases in cells (the combination of variable and group
) and is the expected (theoretical) frequency based on similar proportions of positive
answers among groups (Pearson, 1900; Conover, 1999). Kendall´s Tau test will be used to
analyze the existence of an order among categories in terms of the probability of a positive
response.
Similar to the JT statistics, Kendall´s Tau is a directional statistic and it tests and
, being both variables ordinals (Kendall, 1955; Brown and Benedetti,
1977). In the case of Link and ER, it tests how the ordering of the strategies follows
increasing probabilities of a positive value. Since the associations to test has a rectangular
form (8 strategies x 2 possible values of Link and ER), Kendall´s Tau-c was used. The
Kendall´s Tau-c ( ) statistic ranges from 0 (no association) up to 100 (perfect association)
and it is estimated as:
. Where is the minimum number of non-tied cases,
the number of concordant pairs, is the number of discordant pairs, is the total sample.
2.1.4. The multivariate analysis of variances (MANOVA)
The general notation of MANOVA is: . Where total variability
( ) is a matrix (p=number of explicative variables) whose variability is explained by
the sum of variability due to the strategy ( ), the variability due to the year ( ), the
variability due to the interaction of both factors ( ) and the error variability ( ).
The dependent variables to test is the innovative structure (innovation expenditure and
distribution, qualified human resources, linkages and access to external funds) controlled by
size, sector and capital ownership, in a multifactor design where the strategy s (C, N, SP
and NI) at the time t (t-2/t-1 and t-1/t) are the factors (independent variables). MANOVA
tests the within and between groups’ variability (variances and covariances). The null
hypothesis states that scores on the variables from each one of the groups are sampled
Appendix
- 269 -
from a normal distribution means vector. Then, MANOVA will be used to test if there is
an underlying vector of means for each combination of past and present strategies.
Standard statistical software estimates four MANOVA statistics: Wilks’ lambda, Pillai’s
trace, Lawley–Hotelling trace, and Roy’s largest root. The four statistics tests are performed
on , where H is the sum of squares and cross products matrix and E denote
error sums of squares and cross products matrix. Being the eigenvalue of the matrix
A, the four statistics are formally written as follows:
[ ] ∑
[ ] ∑
∏
2.2. Non-parametric estimations - Results
2.2.1.a Innovative strategies – Number of firms (Pearson’s std. residuals)
Sub-period
Strategies
1998–2001 2002–2004 2005–2006
Innovative Continuous Continuous
446
366 (11.2)
325 (8.5)
Sporadic
41 (2.0)
Sporadic New
80 (5.2)
33 (7.3)
NI
47 (6.5)
Non-Innovative New Continuous
145 (10.0)
120 (4.4)
Sporadic
354 25 (3.8)
NI New
209 (11.8)
67 (8.0)
NI
142 (13.2) Source: own elaboration based on INDEC (2010).
2.2.1.b. Pearson´s chi-square
(a) Likelihood ratio
791.96* 1074.91*
811.25* 1051.9* Obs.:800. *significant at 99% level. a: Assuming the alternative hypothesis. Degrees of freedom=3. Source: own elaboration based on INDEC (2010).
Appendix
- 270 -
2.2.1. Innovation expenditure frequency: number of years (average ranges)
IS N 1998-2001 2002-2004 2005-2006
NI-NI 142 179.39 145.00 129.00
NI-N 67 176.50 145.00 445.23
SP-NI 47 507.27 145.00 129.00
SP-N 33 521.38 145.00 431.82
N-SP 25 176.50 467.16 129.00
N-C 120 176.50 533.79 541.96
C-SP 41 551.79 436.39 129.00
C-C 325 596.43 568.83 548.89
KW (DF=7) 686.31*** 643.02*** 647.41***
Observed JT 196028.50 203301.50 186083.00
Typified JT 21.72*** 23.72*** 19.33***
KW (DF=1) 308.16*** 7.70*** 0.398
Observed JT 39000.00 2249.50 19997.50
Typified JT 17.55*** 2.77*** 0.631 Obs.: 800. **, * significant at 99%. Columns correspond to the each ENIT. KW: Kruskal-Wallis statistic. JT:
Jonckheere-Terpstraa. combination of strategies. . DF: degrees of freedom. Source: own elaboration based on INDEC (2010).
2.2.2. Innovation Intensity: total innovation expenditures to sales (average ranges)
IS N
NI-NI 142 180 145 128 180 145 128
NI-N 67 177 145 462 177 145 457
SP-NI 47 511 145 128 503 145 128
SP-N 33 563 145 510 567 145 502
N-SP 25 177 540 128 177 525 128
N-C 120 177 539 554 177 534 548
C-SP 41 552 498 128 557 477 128
C-C 325 591 554 534 592 559 537
KW (DF=7) 650.685** 582.807** 545.587** 651.697** 585.908** 546.060**
Observed JT 194339 198520 181080 194471 200296 181830
Typified JT 21** 21** 16** 21** 22** 17**
KW (DF=3) 319.949** 5.453 173.506** 319.591** 12.690** 172.741**
Observed JT 62388 36948 43051 62261 38724 43547
Typified JT 17.022** 1.367 5.082** 16.944** 2.447* 5.453**
Obs.: 800. **, * significant at 99%, 95% level. Innovation expenditure to sales. Annual average of the sub-period (in respect of average value of the sector at 2-digit ISIC). t-2: 1998-2001; t-1: 2002-2004; t:
2005-2006. KW: Kruskal-Wallis statistic. JT: Jonckheere-Terpstraa. combination of strategies.
. DF: degrees of freedom. Source: own elaboration based on INDEC (2010).
Appendix
- 271 -
2.2.3 Innovation balance: distribution of innovation expenditure (average ranges)
IS N
NI-NI 142 184 145 128 184 145 128
NI-N 67 179 145 438 179 145 434
SP-NI 47 515 145 128 507 145 128
SP-N 33 522 145 511 506 145 493
N-SP 25 179 445 128 179 434 128
N-C 120 179 538 517 179 537 520
C-SP 41 555 454 128 546 434 128
C-C 325 591 567 552 595 571 554
KW (DF=7) 637.564* 595.209* 552.777* 641.915* 604.534* 556.616*
Observed JT 194155 203171 186718 195424 204216 187482
Typified JT 20.72* 22.81* 18.03* 21.09* 23.18* 18.27*
Obs.: 800. *significant at 99% level. : innovation balance index (in respect of average value of the sector, 2-digit ISIC). t-2: 1998-2001; t-1: 2002-2004; t: 2005-2006. KW: Kruskal-Wallis statistic. JT: Jonckheere-
Terpstraa. combination of strategies. DF: Degrees of freedom. Source: own elaboration based on INDEC (2010).
2.2.4. Qualified human resources: professionals to total employment
IS N
NI-NI 142 265 305 291 252 292 259
NI-N 67 366 364 392 348 355 363
SP-NI 47 384 324 402 379 319 399
SP-N 33 467 459 497 426 457 463
N-SP 25 237 315 353 253 313 341
N-C 120 373 427 412 381 437 432
C-SP 41 384 369 366 369 353 346
C-C 325 487 455 444 499 462 464
KW (DF=7) 117.490* 60.221* 51.591* 139.098* 77.049* 88.240*
Observed JT 157005.5 146408.0 142823.0 161905.5 149532.5 151770.0
Typified JT 9.695* 6.786* 5.632* 11.064* 7.662* 8.101*
Obs.: 800. *significant at 99% level. skilled human resources to total employment, annual average of the sub-period (in respect of average value of the sector at 2-digit ISIC). t-2: 1998-2001; t-1: 2002-2004; t: 2005-2006. KW: Kruskal-Wallis statistic. JT: Jonckheere-Terpstraa. DF: degrees of freedom. Source: own elaboration based on INDEC (2010).
2.2.5.a. Linages with the national innovation system – probability of interaction
Strategy
Total NI-NI NI-N SP-NI SP-N N-SP N-C C-SP C-C
0 Observed 102 43 11 6 21 57 9 48 297 Expected 52.7 24.9 17.4 12.3 9.3 44.6 15.2 120.7 297.0 Std. Res. 6.8 3.6 -1.5 -1.8 3.8 1.9 -1.6 -6.6 0.0
1 Observed 40 24 36 27 4 63 32 277 503 Expected 89.3 42.1 29.6 20.7 15.7 75.5 25.8 204.3 503.0 Std. Res. -5.2 -2.8 1.2 1.4 -3.0 -1.4 1.2 5.1 0.0
0 Observed 125 38 41 19 20 42 35 99 419 Expected 74.4 35.1 24.6 17.3 13.1 62.9 21.5 170.2 419.0 Std. Res 5.9 0.5 3.3 0.4 1.9 -2.6 2.,9 -5.5 0.0
1 Observed 17 29 6 14 5 78 6 226 381 Expected 67.6 31.9 22.4 15.7 11.9 57.2 19.5 154.8 381.0 Std. Res -6.2 -0.5 -3.5 -0.4 -2.0 2.8 -3.1 5.7 0.0
Total Observed 142 67 47 33 25 120 41 325 800 Expected 142 67 47 33 25 120 41 325 800
at least one linkage in the sub-period, average probability. t-2: 1998-2001; t: 2005-2006. Source: own elaboration based on INDEC (2010).
Appendix
- 272 -
2.2.5.b. Pearson´s chi-square and Kendal´s tau-c
Pearson´s chi-square (DF=7) Kendal´s tau-c
(a) Likelihood ratio Std. Err. (a) T app. (b)
205.843* 213.339* 0.468* 0.034 13.949
198.755*a 216.511* 0.457* 0.034 13.503 Obs.:800. *significant at 99% level. a: Assuming the alternative hypothesis. b. Std. Errors based on the null hypothesis. DF: degrees of freedom. Source: own elaboration based on INDEC (2010).
2.2.6.a. Access to external funds – probability of accessing
Strategy
Total NI-NI NI-N SP-NI SP-N N-SP N-C C-SP C-C
0
Observed 141 67 31 20 25 120 28 211 643
Expected 114.1 53.9 37.8 26.5 20.1 96.5 33.0 261.2 643.0
Std. Res. 2.5 1.8 -1.1 -1.3 1.1 2.4 -0.9 -3.1 0.0
1
Observed 1 0 16 13 0 0 13 114 157
Expected 27.9 13.1 9.2 6.5 4.9 23.6 8.0 63.8 157.0
Std. Res. -5.1 -3.6 2.2 2.6 -2.2 -4.9 1.7 6.3 0.0
0
Observed 142 66 47 32 20 101 38 287 733
Expected 130.1 61.4 43.1 30.2 22.9 110.0 37.6 297.8 733.0
Std. Res. 1.0 0.6 0.6 0.3 -0.6 -0.9 0.1 -0.6 0.0
1
Observed 0 1 0 1 5 19 3 38 67
Expected 11.9 5.6 3.9 2.8 2.1 10.1 3.4 27.2 67.0
Std. Res. -3.4 -1.9 -2.0 -1.1 2.0 2.8 -0.2 2.1 0.0
0
Observed 142 61 47 26 25 103 41 252 697
Expected 123.7 58.4 40.9 28.8 21.8 104.6 35.7 283.2 697.0
Std. Res 1.6 0.3 0.9 -0.5 0.7 -0.2 0.9 -1.9 0.0
1
Observed 0 6 0 7 0 17 0 73 103
Expected 18.3 8.6 6.1 4.2 3.2 15.5 5.3 41.8 103.0
Std. Res -4.3 -0.9 -2.5 1.3 -1.8 0.4 -2.3 4.8 0.0
Total Observed 142 142 67 47 33 25 120 41 325
Expected 142 142 67 47 33 25 120 41 325
At least one external source of funding in the sub-period. t-2: 1998-2001; t-1: 2002-2004; t: 2005-2006. Source: own elaboration based on INDEC (2010).
2.2.6.b.Pearson´s chi-square and Kendal´s tau-c
Pearson´s chi-square (DF=7) Kendal´s tau-c
(a) Likelihood ratio Std. Err. (a) T app. (b)
151.35* 203.45* 0.289* 0.028 10.327
40.46* 55.35* 0.098* 0.018 5.436
67.45* 95.7373* 0.183* 0.024 7.676 Obs.:800. *significant at 99% level. a: Assuming the alternative hypothesis. b. Std. Errors based on the null hypothesis. DF: degrees of freedom. Source: own elaboration based on INDEC (2010).
Appendix
- 273 -
2.3. Multivariate analysis of variances - Marginal effects
II IB QHR Link ER
Mar. Eff. Std. Err.
Mar. Eff. Std. Err.
Mar. Eff. Std. Err.
Mar. Eff. Std. Err.
Mar. Eff.
Std. Err.
NI-NI 0.000 0.003 0.000 0.009 0.042*** 0.013 0.294*** 0.023 0.002 0.016
NI-N 0.007 0.005 0.118*** 0.014 0.076*** 0.019 0.573*** 0.035 0.051** 0.025
SP-NI 0.025*** 0.005 0.150*** 0.014 0.113*** 0.020 0.614*** 0.035 0.083*** 0.025
SP-N 0.031*** 0.005 0.330*** 0.016 0.099*** 0.021 0.831*** 0.038 0.162*** 0.027
N-SP 0.000 0.011 0.000 0.035 0.053 0.048 0.172** 0.087 0.002 0.062
N-C 0.025*** 0.006 0.342*** 0.018 0.092*** 0.025 0.537*** 0.045 0.144*** 0.032
C-SP 0.000 0.009 0.000 0.028 0.057 0.038 0.793*** 0.070 0.002 0.049
C-C 0.027*** 0.004 0.385*** 0.013 0.106*** 0.018 0.865*** 0.033 0.227*** 0.023
t-1 0.024*** 0.003 0.228*** 0.009 0.078*** 0.012 0.648*** 0.022 0.108*** 0.015
T 0.013*** 0.003 0.202*** 0.009 0.092*** 0.012 0.626*** 0.022 0.117*** 0.016
NI-NI t-1 0.000 0.004 0.000 0.012 0.034** 0.016 0.306*** 0.029 0.005 0.021
NI-NI t 0.000 0.005 0.000 0.014 0.049** 0.020 0.120*** 0.038 0.000 0.025
NI-N t-1 0.000 0.006 0.000 0.019 0.079*** 0.026 0.788*** 0.047 0.013 0.033
NI-N t 0.014** 0.007 0.236*** 0.021 0.073** 0.028 0.433*** 0.055 0.090** 0.036
SP-NI t-1 0.050*** 0.005 0.301*** 0.014 0.123*** 0.019 0.462*** 0.035 0.166*** 0.025
SP-NI t 0.000 0.008 0.000 0.025 0.102*** 0.034 0.128* 0.066 0.000 0.043
SP-N t-1 0.042*** 0.003 0.355*** 0.009 0.081** 0.012 0.844*** 0.022 0.112*** 0.016
SP-N t 0.019** 0.010 0.305*** 0.030 0.117** 0.040 0.424*** 0.078 0.212*** 0.052
N-SP t-1 0.000 0.012 0.000 0.039 0.046** 0.053 0.185* 0.096 0.005 0.068
N-SP t 0.000 0.011 0.000 0.034 0.060** 0.047 0.200** 0.090 0.000 0.060
N-C t-1 0.025*** 0.008 0.342*** 0.024 0.085** 0.033 0.550*** 0.060 0.146*** 0.042
N-C t 0.025*** 0.005 0.342*** 0.016 0.099** 0.021 0.650*** 0.041 0.142*** 0.027
C-SP t-1 0.000 0.010 0.000 0.033 0.050** 0.044 0.805*** 0.080 0.005 0.057
C-SP t 0.000 0.009 0.000 0.027 0.064** 0.036 0.146** 0.070 0.000 0.046
C-C t-1 0.027*** 0.007 0.385*** 0.021 0.099** 0.028 0.877*** 0.051 0.229*** 0.036
C-C t 0.027*** 0.003 0.385*** 0.009 0.114** 0.013 0.695*** 0.025 0.225*** 0.017
Obs.: 2400. *, **, ***, significant at 90%, 95% and 99% level. Innovation expenditure to sales. Annual average
of the sub-period. . : Distribution of the total innovation expenditure of the sub-period. Total number
of professionals. Annual average of the sub-period At least one linkage in the sub-period. ER At least one external
source of funding in the sub-period. Control group: t-2. . t-2: 1998-2001; t-1: 2002-2004; t: 2005-2006. Delta metod. Source: own elaboration based on INDEC (2010).
Appendix
- 274 -
2.4. Discriminant analysis
2.4.1. General descriptive statistics and media test (ANOVA)
Mean Std. dev. Min Max Wilks´Lamda F DF1 DF2
t-2 1.043 2.878 0.000 42.180 0.885* 14.752 7 792
t-1 0.971 2.407 0.000 34.730 0.894* 13.408 7 792
t 1.072 2.832 0.000 52.710 0.930* 8.474 7 792
t-2 0.388 0.448 0.000 1.890 0.397* 171.776 7 792
t-1 0.447 0.516 0.000 3.190 0.546* 94.061 7 792
t 0.505 0.548 0.000 3.510 0.575* 83.702 7 792
t-2 0.739 1.050 0.000 10.220 0.944* 6.691 7 792
t-1 0.929 3.203 0.000 83.010 0.986 1.628 7 792
T 0.878 1.024 0.000 9.470 0.971* 3.335 7 792
t-2 0.629 0.483 0.000 1.000 0.743* 39.198 7 792
t 0.476 0.500 0.000 1.000 0.752* 37.402 7 792
t-2 0.196 0.397 0.000 1.000 0.811* 26.400 7 792
t-1 0.084 0.277 0.000 1.000 0.949* 6.027 7 792
t 0.129 0.335 0.000 1.000 0.916* 10.418 7 792
181.904 429.308 5.000 5681.500 0.936* 7.773 7 792
0.159 0.366 0.000 1.000 0.962* 4.511 7 792
Obs.: 800. *, significant at 99% level. : Total employment, average value 1998-2001. : Capital ownership, a firm that has more than 1% of their shares owned by foreign capital in t-1. DF: degrees of freedom. Source: own elaboration based on INDEC (2010).
2.4.2. Pooled within-groups correlation matrix
t-2 t-1 T t-2 t-1 T t-2 t-1 T t-2 T t-2 t-1 t
t-2 1.00
t-1 0.20 1.00
T 0.07 0.09 1.00
t-2 -0.09 -0.01 -0.01 1.00
t-1 -0.04 0.01 0.01 0.19 1.00
T -0.08 -0.04 -0.08 0.17 0.26 1.00
t-2 -0.04 0.02 0.01 0.20 0.11 0.03 1.00
t-1 -0.02 0.17 0.03 0.05 0.05 0.01 0.34 1.00
T -0.03 0.04 0.01 0.16 0.08 0.03 0.43 0.41 1.00
t-2 0.03 0.01 -0.01 0.09 0.04 0.01 0.10 0.06 0.08 1.00
T 0.00 0.02 0.03 0.03 0.18 0.17 0.07 0.05 0.15 0.09 1.00
t-2 0.14 0.04 -0.01 0.00 0.02 0.01 -0.03 0.01 0.00 0.06 -0.05 1.00
t-1 0.01 0.08 0.07 0.03 0.06 0.01 0.05 0.12 0.07 0.01 0.06 0.11 1.00
T -0.01 0.00 0.11 0.05 0.04 0.04 0.07 0.09 0.03 0.03 0.05 0.05 0.32 1.00
-0.05 0.02 -0.02 0.05 0.06 -0.02 0.06 0.03 0.05 0.04 0.02 0.05 0.00 -0.02 1.00
-0.01 -0.01 0.02 -0.02 0.05 0.03 0.20 0.06 0.19 0.06 0.14 -0.05 -0.04 -0.03 0.16 1.00
Bold: significant at 95%. Obs.: 800. Source: own elaboration based on INDEC (2010).
Appendix
- 275 -
2.4.3. Pooled within-groups covariance matrix
t-2 t-1 T t-2 t-1 t t-2 t-1 T t-2 t t-2 t-1 t
t-2 7.39
t-1 1.23 5.23
T 0.51 0.54 7.53
t-2 -0.07 -0.01 -0.01 0.08
t-1 -0.04 0.01 0.01 0.02 0.15
T -0.10 -0.04 -0.09 0.02 0.04 0.17
t-2 -0.12 0.06 0.04 0.06 0.05 0.01 1.05
t-1 -0.15 1.21 0.24 0.05 0.06 0.01 1.12 10.21
T -0.08 0.10 0.03 0.05 0.03 0.01 0.45 1.34 1.03
t-2 0.03 0.01 -0.01 0.01 0.01 0.00 0.04 0.08 0.03 0.18
T 0.00 0.02 0.04 0.00 0.03 0.03 0.03 0.07 0.07 0.02 0.19
t-2 0.14 0.03 -0.01 0.00 0.00 0.00 -0.01 0.01 0.00 0.01 -0.01 0.13
t-1 0.01 0.05 0.05 0.00 0.01 0.00 0.01 0.11 0.02 0.00 0.01 0.01 0.07
T -0.01 0.00 0.10 0.00 0.00 0.01 0.02 0.09 0.01 0.00 0.01 0.01 0.03 0.10
-57.62 19.13 -26.37 5.49 9.25 -3.80 27.33 39.34 20.5 6.14 4.27 6.89 -0.40 -2.75 173981
0.00 -0.01 0.02 0.00 0.01 0.00 0.08 0.07 0.07 0.01 0.02 -0.01 0.00 0.00 23.45 0.13 Obs.: 800. Source: own elaboration based on INDEC (2010).
2.4.4. Discriminant functions
Eigenvalue Acc. Var. (%) Canonic corr. Contrasts Wilks´Lamda Chi-square DF
1 2.458 65.0 0.843 1 to 7 0.105* 1772.685 112
2 0.914 89.2 0.691 2 to 7 0.364* 796.221 90
3 0.329 97.9 0.498 3 to 7 0.696* 285.396 70
4 0.045 99.1 0.207 4 to 7 0.925 61.332 52
5 0.017 99.5 0.128 5 to 7 0.966 26.973 36
6 0.010 99.8 0.099 6 to 7 0.983 13.883 22
7 0.008 100.0 0.088 7 0.992 6.137 10 Obs.: 800. *, significant at 99% level. Source: own elaboration based on INDEC (2010).
2.4.5. Estimated and observed cluster membership
Observed Estimated
Total NINI NIN SPNI SPN NSP NC CSP CC
Number of cases
NINI 139 2 1 0 0 0 0 0 142
NIN 25 42 0 0 0 0 0 0 67
SPNI 18 0 29 0 0 0 0 0 47
SPN 6 4 4 11 0 0 0 8 33
NSP 16 0 0 0 6 3 0 0 25
NC 17 10 0 0 2 91 0 0 120
CSP 10 0 4 0 1 0 22 4 41
CC 8 3 4 4 0 24 2 280 325
Source: own elaboration based on INDEC (2010).
Appendix
- 276 -
2.4.6. Mahalanobis square distance (Ms-d) - Between groups square distance
IS observed IS Estimated
NI-NI NI-N SP-NI SP-N N-SP N-C C-SP C-C
NI-NI Ms-d 0.0000
F 0.0000
NI-N Ms-d 2.3718 0.0000
F 6.6201 0.0000
SP-NI Ms-d 5.3889 7.9549 0.0000
F 11.6681 13.4735 0.0000
SP-N Ms-d 9.7023 6.8116 3.9110 0.0000
F 15.9299 9.2345 4.6493 0.0000
N-SP Ms-d 3.2155 6.6743 10.3217 16.1401 0.0000
F 4.1912 7.4508 10.3284 14.0770 0.0000
N-C Ms-d 7.2675 4.5293 13.8461 12.0809 5.3058 0.0000
F 28.9822 11.9404 28.6724 19.1726 6.7310 0.0000
C-SP Ms-d 8.5873 12.6429 2.5667 7.5901 9.2350 13.2852 0.0000
F 16.7514 19.7177 3.4463 8.5092 8.7941 24.8934 0.0000
C-C Ms-d 14.9413 12.5645 8.5365 5.3282 15.3779 9.4381 5.8082 0.0000
F 90.5354 42.7950 21.4928 9.7875 21.8892 50.7181 12.9660 0.0000 Obs.: 800. All Ms-d significantly different at 99% (p-value=0). F with 16 and 777 degrees of freedom. Source: own elaboration based on INDEC (2010).
2.4.7 Innovative strategies – distribution by sector and size
ISIC Continuous New Sporadic NI
Small Medium Large Total Small Medium Large Total Small Medium Large Total Small Medium Large Total
15 24 17 12 53 15 6 0 21 35 4 2 41 22 1 0 23
16 0 0 1 1 1 0 1 2 0 0 0 0 0 0 0 0
17 18 7 0 25 5 0 0 5 21 0 0 21 27 0 0 27
18 5 0 0 5 5 0 0 5 9 0 0 9 8 0 0 8
19 4 4 1 9 2 1 0 3 12 0 0 12 2 0 0 2
20 4 1 0 5 3 1 0 4 8 0 0 8 6 0 0 6
21 4 0 1 5 2 1 1 4 7 1 0 8 7 0 0 7
22 7 6 1 14 7 0 0 7 15 2 0 17 5 0 0 5
23 0 2 1 3 0 0 3 3 0 1 0 1 0 0 0 0
24 19 16 7 42 9 0 1 10 4 3 2 9 5 0 0 5
25 15 6 1 22 6 0 1 7 8 0 0 8 3 0 0 3
26 14 5 1 20 4 0 0 4 16 0 0 16 7 0 0 7
27 8 2 2 12 2 0 1 3 4 1 0 5 4 0 0 4
28 10 3 0 13 6 0 0 6 13 0 0 13 10 0 0 10
29 35 4 1 40 12 2 0 14 14 1 0 15 4 0 0 4
30 1 0 0 1 0 0 0 0 0 0 0 0 1 0 0 1
31 17 0 0 17 4 0 0 4 4 0 0 4 5 1 0 6
32 0 4 0 4 3 0 0 3 2 0 0 2 0 0 0 0
33 5 0 0 5 2 0 0 2 2 0 0 2 3 0 0 3
34 14 1 3 18 6 1 0 7 7 0 0 7 5 0 0 5
35 2 0 0 2 2 0 0 2 3 0 0 3 5 0 0 5
36 7 2 0 9 4 0 0 4 11 1 0 12 11 0 0 11
Total 213 80 32 325 100 12 8 120 195 14 4 213 140 2 0 142
Obs.: 800. Continuous: C-C; New: N-C; Non-innovative: NI-NI; Sporadic: rest of the firms. ISIC: rev. 3.1. Small firms, sales < AR$25 mill.; medium firms, AR$25 mill. < sales < AR$ 100 mill.; large firms, sales > AR$ 100 mill; in 1998-2001. Source: own elaboration based on INDEC (2010).
Appendix
- 277 -
Appendix 3: Innovative strategies, persistence and performance
3.1. Non-parametric estimations
3.1.1.a. Innovation: product, process, organizational or commercial innovations (probabilities of positive responses)
Strategy Total
NI-NI NI-N SP-NI SP-N N-SP N-C C-SP C-C
0
Observed 128 57 7 5 24 82 8 19 330
Expected 58.6 27.6 19.4 13.6 10.3 49.5 16.9 134.1 330
Std. Res. 9.1 5.6 -2.8 -2.3 4.3 4.6 -2.2 -9.9
1
Observed 14 10 40 28 1 38 33 306 470
Expected 83.4 39.4 27.6 19.4 14.7 70.5 24.1 190.9 470
Std. Res. -7.6 -4.7 2.4 2.0 -3.6 -3.9 1.8 8.3
0
Observed 141 35 31 19 14 54 20 168 482
Expected 85.6 40.4 28.3 19.9 15.1 72.3 24.7 195.8 482
Std. Res. 6.0 -0.8 0.5 -0.2 -0.3 -2.2 -0.9 -2.0
1
Observed 1 32 16 14 11 66 21 157 318
Expected 56.4 26.6 18.7 13.1 9.9 47.7 16.3 129.2 318
Std. Res. -7.4 1.0 -0.6 0.2 0.3 2.6 1.2 2.4
0
Observed 142 32 45 12 24 28 40 60 383
Expected 68.0 32.1 22.5 15.8 12.0 57.5 19.6 155.6 383
Std. Res 9.0 0.0 4.7 -1.0 3.5 -3.9 4.6 -7.7
1
Observed 0 35 2 21 1 92 1 265 417
Expected 74.0 34.9 24.5 17.2 13.0 62.6 21.4 169.4 417
Std. Res -8.6 0.0 -4.5 0.9 -3.3 3.7 -4.4 7.3
Total Observed 142 67 47 33 25 120 41 325 800
Expected 142 67 47 33 25 120 41 325 800
Obs.: 800. At least one innovation in the sub-period. t-2: 1998-2001; t-1: 2002-2004; t: 2005-2006. Source: own elaboration based on INDEC (2010).
3.1.1.b. Pearson´s chi-square and Kendal´s tau-c
Pearson´s chi-square (DF=7) Kendal´s tau-c
(a) Likelihood ratio Std. Err. (a) T app. (b)
459.24* 525.39* 0.70333* 0.02620 26.84324
116.96* 158.82* 0.29568* 0.03475 8.50938
404.91* 495.95* 0.62039* 0.02940 21.09989 Obs.:800. *significant at 99% level. a: Assuming the alternative hypothesis. b. Std. Errors based on the null hypothesis. DF: degrees of freedom. Source: own elaboration based on INDEC (2010).
Appendix
- 278 -
3.1.2. Productivity and innovative strategies (average ranges)
N
NI-NI 142 292 275 273 262 245 242
NI-N 67 372 403 400 363 396 393
SP-NI 47 411 395 379 400 382 363
SP-N 33 380 347 400 379 336 383
N-SP 25 394 386 362 338 337 308
N-C 120 396 409 397 413 423 414
C-SP 41 397 373 366 399 371 363
C-C 325 457 462 468 472 479 485
KW (DF=7) 52.2** 67.4** 73.7** 86.2** 107.7** 117.5**
Observed JT 147316.0 149691.0 151363.0 154342.0 157275.0 159233.0
Typified JT 6.9** 7.5** 8.0** 8.8** 9.6** 10.2**
KW (DF=3) 9.3* 10.7* 17.2** 14.3** 18.2** 26.4**
Observed JT 39590.0 39554.0 41138.0 40503.0 40823.0 42335.0
Typified JT 3.0** 3.0** 3.9** 3.5** 3.7** 4.6**
Obs.: 800. *, ** significant at 95% and 99% level. labor productivity, sales to total employment. Annual average of the sub-period (in respect of average value of the sector at 2-digit ISIC classification). t-2: 1998-2001; t-1: 2002-2004; t: 2005-2006. KW: Kruskal-Wallis statistic. JT: Jonckheere-
Terpstraa. categories,combination of strategies. . DF: degrees of freedom. Source: own elaboration based on INDEC (2010).
3.1.3. Growth and innovative strategies – Total Panel (average ranges)
N
NI-NI 280 263 251 253 239 228 280
NI-N 312 310 301 308 307 296 312
SP-NI 345 321 326 368 335 341 345
SP-N 341 328 330 349 326 333 341
N-SP 256 267 263 249 255 255 256
N-C 402 407 426 414 419 436 402
C-SP 359 375 368 388 401 386 359
C-C 501 509 509 502 512 513 501
KW (DF=7) 125.7* 149.9* 162.9* 145.7* 173.3* 188.3*
Observed JT 162257.5 166669.5 168367.0 165435.0 169891.5 171570.0
Typified JT 11.0* 12.2* 12.7* 11.9* 13.1* 13.6*
KW (DF=6) 25.4* 34.7* 48.9* 44.5* 52.6* 67.7*
Observed JT 52393.0 54208.5 55811.0 55139.0 56559.5 57910.5
Typified JT 4.2* 5.3* 6.3* 5.9* 6.7* 7.5*
Obs.: 800. *significant at 99% level. total employment. Annual average of the sub-period (in respect of average value of the sector at 2-digit ISIC classification). t-2: 1998-2001; t-1: 2002-2004; t: 2005-
2006. KW: Kruskal-Wallis statistic. JT: Jonckheere-Terpstraa. categories,combination of strategies.
. DF: degrees of freedom. Source: own elaboration based on INDEC (2010).
Appendix
- 279 -
3.1.4. Growth and innovative strategies – SMEs (average ranges)
N
NI-NI 142 278 262 250 252 239 228
NI-N 67 309 308 299 307 306 296
SP-NI 46 337 314 321 360 328 335
SP-N 32 333 319 322 337 314 322
N-SP 25 255 265 262 249 255 255
N-C 112 385 385 405 391 394 413
C-SP 39 344 361 353 371 385 369
C-C 293 467 477 477 471 482 482
KW (DF=7) 97.6* 121.0* 132.7* 117.4* 144.8* 158.3*
Observed JT 142902.5 147352.0 148916.5 146219.5 150836.0 152305.0
Typified JT 9.6* 10.9* 11.4* 10.6* 12.0* 12.4*
Obs.: 756. *significant at 99% level. total employment, annual average of the sub-period (in respect of average value of the sector, 2-digit ISIC classification). t-2: 1998-2001; t-1: 2002-2004; t: 2005-2006. SMEs: sales<AR$200 million in 1998-2001. KW: Kruskal-Wallis statistic. JT: Jonckheere-Terpstraa. DF: Degrees of freedom. Source: own elaboration based on INDEC (2010).
3.1.5.a. Exporters (probabilities of exports)
Strategy
Total NI-NI NI-N SP-NI SP-N N-SP N-C C-SP C-C
0
Observed 83 34 18 13 12 32 13 54 259
Expected 46.0 21.7 15.2 10.7 8.1 38.9 13.3 105.2 259
Std. Res. 5.5 2.6 0.7 0.7 1.4 -1.1 -0.1 -5.0
1
Observed 59 33 29 20 13 88 28 271 541
Expected 96.0 45.3 31.8 22.3 16.9 81.2 27.7 219.8 541
Std. Res. -3.8 -1.8 -0.5 -0.5 -1.0 0.8 0.1 3.5
0
Observed 88 28 23 11 14 32 14 62 272
Expected 48.3 22.8 16.0 11.2 8.5 40.8 13.9 110.5 272
Std. Res. 5.7 1.1 1.8 -0.1 1.9 -1.4 0.0 -4.6
1
Observed 54 39 24 22 11 88 27 263 528
Expected 93.7 44.2 31.0 21.8 16.5 79.2 27.1 214.5 528
Std. Res. -4.1 -0.8 -1.3 0.0 -1.4 1.0 0.0 3.3
0
Observed 108 39 26 17 16 40 18 83 347
Expected 61.6 29.1 20.4 14.3 10.8 52.1 17.8 141.0 347
Std. Res 5.9 1.8 1.2 0.7 1.6 -1.7 0.1 -4.9
1
Observed 34 28 21 16 9 80 23 242 453
Expected 80.4 37.9 26.6 18.7 14.2 68.0 23.2 184.0 453
Std. Res -5.2 -1.6 -1.1 -0.6 -1.4 1.5 0.0 4.3
Total Observed 142 67 47 33 25 120 41 325 800
Expected 142 67 47 33 25 120 41 325 800
Obs.: 800. =1 if exports>0 during the sub-period, percentage of firms. t-2: 1998-2001; t-1: 2002-2004; t: 2005-2006. Source: own elaboration based on INDEC (2010).
3.1.5.b. Pearson´s chi-square and Kendal´s tau-c
Pearson´s chi-square (DF=7) Kendal´s tau-c
(a) Likelihood ratio Std. Err. (a) T app. (b)
97.375* 97.66* 0.35116* 0.03515 9.98956
96.525* 95.92* 0.34020* 0.03577 9.51049
122.732* 126.25* 0.40916* 0.03563 11.48344 Obs.:800. *significant at 99% level. a: Assuming the alternative hypothesis. b. Std. Errors based on the null hypothesis. DF: degrees of freedom. Source: own elaboration based on INDEC (2010).
Appendix
- 280 -
3.1.6. Total exports and innovative strategies (average ranges)
N
NI-NI 142 274 262 256 263 249 251
NI-N 67 291 314 324 295 313 328
SP-NI 47 364 332 340 371 337 341
SP-N 33 382 367 370 377 368 355
N-SP 25 295 291 284 299 296 289
N-C 120 417 430 428 423 436 436
C-SP 41 401 413 411 398 411 404
C-C 325 487 488 489 489 491 489
KW (DF=7) 114.10* 124.70* 133.29* 122.81* 137.54* 138.18*
Observed JT 159775 161529 161789 160772 163202 162090
Typified JT 10.48* 11.00* 11.33* 10.76* 11.47* 11.41*
KW (DF=3) 25.27* 22.78* 26.21* 23.88* 22.01* 24.09*
Observed JT 42258 41537 42002 41875 41346 41437
Typified JT 4.62* 4.18* 4.50* 4.38* 4.07* 4.16*
Obs.:800. * significant at 99% level. X exports to sales, average annual values for each sub-period (in respect of average value of the sector, 2-digit ISIC). t-2: 1998-2001; t-1: 2002-2004; t: 2005-2006.
KW: Kruskal-Wallis statistic. JT: Jonckheere-Terpstraa. combination of strategies. . DF: degrees of freedom. Source: own elaboration based on INDEC (2010)
Appendix
- 281 -
3.2. Persistence of innovation (probit and zero inflated models)
Dep. Var.: Inno RE 1 RE 2 RE 3 RE 4 ZI
Coeff. S.E. Coeff. S.E. Coeff. S.E. Coeff. S.E. Coeff. S.E.
Structural equation
-0.4959*** 0.1865 -0.0353 0.1935 -0.4320 0.3065 -0.0474 0.3838 0.5795 0.4147
1.4761*** 0.3068 2.2978*** 0.44915 1.83355*** 0.19594
1.2677*** 0.19364 1.27017*** 0.23710 1.58197*** 0.20654
0.7090*** 0.24763 0.51590 0.38781 0.41555 0.42659
* -1.2211*** 0.4389 -0.7517*** 0.55809
* -0.04023 0.40881 -0.30851 0.43660
* 0.01185 0.5030 0.11628 0.41570
1.31278 1.03342 0.36659 1.03742 0.68116 1.21462 0.14065 0.25781
1.09379*** 0.28848 0.34510 0.34193 0.42468 0.38399 0.10951 0.15800
-0.48655 0.39077 -0.35284 0.42410 -0.50997 0.48783 -0.0375 0.05842
1.061838*** 0.13938 1.04181*** 0.15503 1.150139* 0.17081 0.49944 0.07247
0.358550** 0.13167 0.29331** 0.14483 0.2916018* 0.16489 0.08752 0.05335
15 - Food and Beverage 0.29331 0.25820 0.05618 0.19853 0.05362 0.23569 0.05650 0.28344 0.00202 0.13346
16 – Tobacco 0.39144 0.82770 0.57092 0.65931 0.39628 0.73580 0.54816 0.88261 0.10982 0.16603
17 – Textiles -0.15323 0.27708 -0.19307 0.21508 -0.15232 0.25626 -0.21762 0.30953 -0.1270 0.14984
18 - Wearing apparel; dres. & fur dyeing -0.04005 0.34835 -0.10131 0.27104 -0.15391 0.32245 -0.14885 0.38660 -0.1211 0.19193
19 – Leath.; lug., hand., harness & foot. 0.15448 0.34443 -0.01102 0.26955 -0.01352 0.31656 -0.01947 0.38036 -0.0517 0.18170
20 - Wood & wood & cork pds.(e/ furn) 0.25743 0.36203 0.07577 0.28302 0.15889 0.33214 0.09906 0.39897 -0.0102 0.20612
21 - Paper and paper products -0.42216 0.37519 -0.44926 0.29290 -0.43651 0.34735 -0.52308 0.42039 -0.3344 0.23873
22 – Publ., print. & repro. of rec. media -0.05078 0.30496 -0.06487 0.23768 -0.12213 0.28092 -0.17260 0.33955 -0.1128 0.16250
23 - Coke, ref. petr. pds.& nuclear fuel 0.72839 0.56567 0.12894 0.43121 0.21880 0.49913 0.21834 0.59673 0.08332 0.18346
24 - Chemicals 0.514070* 0.28677 -0.04317 0.22473 -0.09571 0.26382 -0.14024 0.31800 -0.0405 0.14148
25 - Rubber and plastics products 0.49798 0.31441 0.09208 0.23989 0.04741 0.28094 0.05166 0.34015 0.00195 0.15813
26 - Other non-metallic mineral pds. 0.25793 0.30369 -0.05838 0.23642 -0.03744 0.27907 -0.05327 0.33581 -0.0310 0.16260
27 - Basic metals 0.12242 0.35783 -0.18730 0.28224 -0.10381 0.33072 -0.21741 0.40217 -0.1189 0.19611
28 – Fabric. metal pds. (e/mach.& eq.) 0.22487 0.30903 -0.04904 0.23971 0.04050 0.28343 -0.00554 0.34215 -0.0143 0.15981
29 - Machinery and equipment n.e.c. 0.70151** 0.28427 0.12386 0.21693 0.16342 0.25720 0.17373 0.30926 0.04809 0.13988
30 - Office, acc. & computing mach. -0.49495 1.06040 -0.90083 0.94867 -0.96774 1.12209 -1.28113 1.39173 -0.5193 0.18781
31 – Electr. mach. & apparatus n.e.c. 0.30856 0.33126 -0.15507 0.25823 -0.14500 0.30511 -0.13445 0.36712 -0.0651 0.15399
32 - Radio, TV and comm. eq. 0.98689* 0.50833 0.29059 0.41555 0.35867 0.46580 0.35443 0.55377 0.08020 0.15820
33 – Med., prec. and opt. inst., watches & clocks
0.17717 0.45127 -0.50839 0.36814 -0.42058 0.42926 -0.49726 0.52066 -0.2093 0.25306
34 – M. vehic., trailers & semi-trailers 0.55117* 0.32229 -0.03374 0.25182 -0.00924 0.29583 0.00158 0.35788 -0.0160 0.14719
35 - Other transport equipment -0.55673 0.49574 -0.897675** 0.44331 -0.918405* 0.52139 -1.124531* 0.62545 -0.6898 0.30296
0.00074*** 0.00017 0.00019 0.00060 0.00029 0.00066 0.00039 0.00073 0.00007 0.00016
0.09993 0.13866 -0.16389 0.11570 -0.19446 0.13488 -0.22554 0.16225 -0.0707 0.05949
-0.32665*** 0.07576 0.12374 0.10182 0.252047** 0.13060 0.48898*** 0.16750 0.15810 0.08251
-0.67492*** 0.24199 -1.0385*** 0.20779 -1.72349*** 0.31631 -2.0988*** 0.40163 -2.5495 0.20238
Individual heterogeneity
0.92733*** 0.17365 0.06728 0.15052 0.12909 0.17895 0.12337 0.19312 -0.0773 0.06570
2.99325** 1.49309 1.41562 1.65153 1.75048 1.92774 0.58339 0.71983
0.93487** 0.40845 0.51905 0.47943 0.57810 0.54840 0.36234 0.23713
0.81839 0.74053 0.84930 0.83857 1.09639 0.98565 0.05779 0.10464
0.00012 0.00059 0.00003 0.00064 0.00002 0.00071 0.00005 0.00016
-0.34510 0.36613 -2.42679 1.88607 -1.16827 0.96557 -0.28730 0.61081
0.84152 0.15405 0.29719 0.28026 0.55759 0.26920 0.86619 0.26454
0.41458*** 0.08886 0.08115 0.14064 0.23717 0.17469 0.42866*** 0.14959
Inflated Variable 0.01834 0.02047
Obs.: 2400. ***, ** and *: significance at 99%, 95%, and 90%, respectively. =1 if at least one innovation was achieved
during the sub-period. time-average value of the corresponding variable. omitted due to
collinearity, = 2002–2004; = linkages 2005–2006. Gauss–Hermite estimations, twelve quadrature points. RE: random effect. ZI: zero inflated (test the relationship between those firms which actually expended on innovation, while treating zero values differently, static probit clustered by cases). Source: own elaboration based on INDEC (2010).
Appendix
- 282 -
3.3. Performance and strategies (OLS models)
3.3.1. Strategies and productivity
Dep. Var.: Pd RE1 RE2 RE3 MLE PA
Coeff. S.E. Coeff. S.E. Coeff. S.E. Coeff. S.E. Coeff. S.E.
Structural equation
0.67949*** 0.05689 0.6631*** 0.05629 0.666152*** 0.05638 0.66311*** 0.036181 0.7214227*** 0.03623
0.02205 0.01444 -0.01280 0.01610 -0.01209 0.01608 -0.01280 0.01596 -0.01262 0.01582
0.1252*** 0.03120 0.125959*** 0.03123 0.12521*** 0.03038 0.1224656*** 0.02989
0.1315*** 0.03422 0.129705*** 0.03415 0.13152*** 0.033821 0.1296256*** 0.03374
0.01797 0.03828 0.01680 0.03827 0.017967 0.038997 0.01658 0.03895
15 - Food and Beverage 0.115697** 0.04689 0.11804*** 0.04612 0.11607 0.04584 0.11804** 0.053362 0.1098048** 0.05181
16 – Tobacco 0.10810 0.15779 0.08888 0.14704 0.10480 0.14189 0.0889 0.174926 0.09418 0.17019
17 – Textiles 0.16837*** 0.05367 0.16998*** 0.05365 0.171171*** 0.05322 0.16998*** 0.057061 0.1571685*** 0.05543
18 - Wearing apparel; dres. & fur dyeing -0.0407 0.05616 -0.0480 0.05535 -0.04961 0.05510 -0.04797 0.071249 -0.05029 0.06916
19 – Leath.; lug., hand., harness & foot. 0.139790** 0.06816 0.13628** 0.06766 0.13551** 0.06648 0.13628* 0.072284 0.1294753* 0.07017
20 - Wood & wood & cork pds.(e/ furn) 0.1036 0.08763 0.10409 0.08802 0.10955 0.08705 0.104094 0.074816 0.09830 0.07263
21 - Paper and paper products 0.0974 0.06638 0.09608 0.06478 0.09762 0.06538 0.096079 0.07491 0.08794 0.07274
22 – Publ., print. & repro. of rec. media -0.0639 0.05489 -0.07028 0.05620 -0.06856 0.05594 -0.07028 0.063551 -0.07076 0.06169
23 - Coke, ref. petr. pds.& nuclear fuel 0.211555* 0.11238 0.20539* 0.11501 0.20462 0.11722 0.205389 0.128676 0.20148 0.12491
24 - Chemicals 0.144087*** 0.05183 0.13878*** 0.05143 0.141499*** 0.05162 0.138783** 0.060195 0.1321786** 0.05845
25 - Rubber and plastics products 0.170332*** 0.05452 0.15987*** 0.05403 0.164046*** 0.05346 0.159865** 0.065331 0.1533364** 0.06343
26 - Other non-metallic mineral pds. 0.122002** 0.05193 0.12073** 0.05191 0.129106** 0.05175 0.12073* 0.062602 0.1135252* 0.06078
27 - Basic metals 0.33002*** 0.06824 0.32843*** 0.06897 0.326026*** 0.06898 0.328433*** 0.075008 0.3176017*** 0.07283
28 – Fabric. metal pds. (e/mach.& eq.) 0.159203*** 0.06262 0.16202*** 0.06176 0.166415*** 0.06149 0.16202** 0.063867 0.1561935** 0.06200
29 - Machinery and equipment n.e.c. 0.186158*** 0.05504 0.18491*** 0.05488 0.185839*** 0.05471 0.18491*** 0.058552 0.1756403*** 0.05688
30 - Office, acc. & computing mach. 0.11531 0.12930 0.12052 0.15440 0.09992 0.15219 0.120519 0.206632 0.10787 0.20060
31 – Electr. mach. & apparatus n.e.c. 0.118109** 0.05001 0.10848** 0.05053 0.111243** 0.05027 0.108483 0.069104 0.10544 0.06709
32 - Radio, TV and comm. eq. 0.12351 0.09363 0.12507 0.09158 0.13741 0.09041 0.125068 0.106388 0.12516 0.10330
33 – Med., prec. and opt. inst., watches & clocks
0.09186 0.08776 0.08725 0.08401 0.08261 0.08414 0.087246 0.093437 0.08272 0.09070
34 – M. vehic., trailers & semi-trailers 0.230161*** 0.06201 0.22675*** 0.05991 0.236329*** 0.05973 0.22675*** 0.066427 0.2181142*** 0.06450
35 - Other transport equipment 0.08762 0.14027 0.07363 0.13585 0.06949 0.13545 0.07363 0.094574 0.06620 0.09185
0.000001** 0.00000 0.000001** 0.00000 0.000001** 0.00000 0.000001*** 0.00000 0.000001*** 0.00000
0.054897** 0.02646 0.054907** 0.02645 0.0549075* 0.030958 0.052337* 0.03005
-0.10520*** 0.02125 -0.1151*** 0.02114 -0.11469*** 0.02116 -0.11513*** 0.020889 -0.115414*** 0.02144
0.282513*** 0.07289 0.23713*** 0.07397 0.21126*** 0.07187 0.237123*** 0.071103 0.245064*** 0.06917
Individual heterogeneity
0.24652*** 0.06261 0.25395*** 0.06188 0.25856*** 0.06184 0.25395*** 0.038602 0.1960302*** 0.03850
-0.000001* 0.00000 -0.000001* 0.00000 -0.000001* 0.00000 -
0.000001*** 0.00000 -0.000001*** 0.00000
0.00000 0.00000 0.00000 0.00000 0.154919
0.29769 0.29744 0.29744 0.395301 0.006988
0.00000 0.00000 0.00000
Within 0.0033 0.0017 0.0018
Between 0.9133 0.9142 0.9142
Overall 0.8328 0.8353 0.8349
Obs.: 2400. ***, **, and *: significant at 99%, 95%, and 90%, respectively. : sales to employment, average value for the sub-period, in ARS 000, constant prices 1998, natural logarithm.. RE: random effect, robust standard errors. MLE: maximum-likelihood random-effect. PA: GEE population averaged, convergence not
achieved. time-average value of the corresponding variable. omitted due to collinearity, = 2002-2004. Source: own elaboration based on INDEC (2010).
Appendix
- 283 -
3.3.2 Strategies and growth (OLS model)
Dep. Var.: Gr RE 1 RE 2 RE 3 MLE PA
Coeff. S.E. Coeff. S.E. Coeff. S.E. Coeff. S.E. Coeff. S.E.
Structural equation
0.756*** 0.04798 0.74459*** 0.04865 0.7456*** 0.04861 0.7384*** 0.02989 0.0994503*** 0.02406
0.1227*** 0.01739 0.083617*** 0.01779 0.083293*** 0.01761 0.08369*** 0.01790 0.0765282*** 0.01607
0.14108*** 0.03233 0.14232*** 0.03227 0.14154*** 0.03358 0.1462042*** 0.03598
0.06075 0.03843 0.06045 0.03845 0.0605634* 0.03681 0.02248 0.03258
-0.02334 0.04848 -0.02307 0.04848 -0.02319 0.04257 0.00151 0.03634
15 - Food and Beverage 0.11427 0.07603 0.11568 0.07466 0.11608 0.07476 0.116126** 0.05816 0.1654852¨* 0.08593
16 – Tobacco 0.11346 0.09927 0.11440 0.10332 0.12582 0.10660 0.11529 0.19219 0.21604 0.27894
17 – Textiles 0.08263 0.08042 0.07888 0.07943 0.07938 0.07953 0.07902 0.06233 0.08664 0.09195
18 - Wearing apparel; dres. & fur dyeing 0.16391** 0.08147 0.16303 0.07974 0.161069** 0.07976 0.163320** 0.07836 0.18693 0.11575
19 – Leath.; lug., hand., harness & foot. 0.166896* 0.10078 0.1687148* 0.09855 0.1685766* 0.09798 0.16938** 0.07935 0.2427152** 0.11722
20 - Wood & wood & cork pds.(e/ furn) 0.09090 0.10626 0.09818 0.10765 0.10109 0.10742 0.09831 0.08217 0.10982 0.12144
21 - Paper and paper products 0.146838* 0.08408 0.1384998* 0.08209 0.1426873* 0.08181 0.1388348* 0.08161 0.16138 0.12017
22 – Publ., print. & repro. of rec. media 0.10525 0.08223 0.10179 0.08104 0.10425 0.08087 0.10222 0.06973 0.14738 0.10300
23 - Coke, ref. petr. pds.& nuclear fuel 0.03572 0.14083 0.02560 0.14280 0.03393 0.14189 0.02646 0.14016 0.12441 0.20705
24 - Chemicals 0.09677 0.07893 0.08012 0.07765 0.08640 0.07782 0.08066 0.06495 0.15348 0.09572
25 - Rubber and plastics products 0.06041 0.08317 0.04172 0.08197 0.04806 0.08181 0.04216 0.07135 0.10781 0.10513
26 - Other non-metallic mineral pds. 0.06065 0.08241 0.05726 0.08065 0.06505 0.08024 0.05730 0.06853 0.06877 0.10126
27 - Basic metals 0.00453 0.09930 -0.01038 0.09887 -0.00926 0.10018 -0.01057 0.08174 -0.02297 0.12076
28 – Fabric. metal pds. (e/mach.& eq.) 0.10737 0.08024 0.10943 0.07972 0.11419 0.07959 0.10984 0.06996 0.15519 0.10338
29 - Machinery and equipment n.e.c. 0.06029 0.08040 0.05216 0.07899 0.05598 0.07893 0.05217 0.06378 0.07623 0.09346
30 - Office, acc. & computing mach. 0.00901 0.07472 -0.04180 0.07993 -0.04205 0.08064 -0.04386 0.22362 -0.25754 0.33003
31 – Electr. mach. & apparatus n.e.c. 0.09601 0.08430 0.07416 0.08405 0.07833 0.08398 0.07447 0.07567 0.12129 0.11168
32 - Radio, TV and comm. eq. -0.07319 0.11406 -0.07013 0.11379 -0.05481 0.11292 -0.06992 0.11625 -0.02196 0.17093
33 – Med., prec. and opt. inst., watches & clocks
0.03605 0.10695 0.01470 0.11256 0.01387 0.11252 0.01482 0.10247 0.03926 0.15141
34 – M. vehic., trailers & semi-trailers 0.06867 0.08721 0.06089 0.08734 0.06952 0.08798 0.06103 0.07272 0.08899 0.10728
35 - Other transport equipment -0.08536 0.12346 -0.11535 0.12053 -0.11593 0.12041 -0.11633 0.10349 -0.23244 0.15195
0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00001*** 0.00000 0.00001*** 0.00000
0.04624 0.03140 0.04125 0.03185 0.04143 0.03291 0.06203 0.04863
-0.14771*** 0.02076 -0.15255*** 0.02076 -0.15268*** 0.02077 -0.15189*** 0.02312 -0.09218*** 0.01549
0.46483*** 0.08858 0.439682*** 0.09033 0.430398*** 0.08934 0.44041*** 0.06941 0.4933051*** 0.09751
Individual heterogeneity
0.13703*** 0.04569 0.1364945* 0.04595 0.138078*** 0.04595 0.14212*** 0.02980 0.7461159*** 0.02758
0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00001*** 0.00000 0.00000 0.00000
0.00000 0.00000 0.00000 0.03444
0.25384 0.25268 0.25268 0.43121
0.00000 0.00000 0.00000 0.00634
Within 0.00590 0.00530 0.0053
Between 0.93640 0.93670 0.9367
Overall 0.87580 0.87760 0.8775
Obs.: 2400. ***, ** and *: significant at 99%, 95%, and 90%, respectively. : employment, average value for the sub-period, natural logarithm.. RE: random effect, robust standard errors. MLE: maximum-likelihood
random-effect. PA: GEE population averaged. time-average value of the corresponding variable.
omitted due to collinearity, = 2002-2004. Source: own elaboration based on INDEC (2010).
Appendix
- 284 -
3.3.3. Strategies and exports (OLS model)
Dep. Var.: Xpo RE 1 RE 2 RE 3 MLE PA
Coeff. S.E. Coeff. S.E. Coeff. S.E. Coeff. S.E. Coeff. S.E.
Structural equation
0.0831*** 0.01419 0.07894*** 0.01415 0.081060*** 0.01434 0.075643*** 0.01391 0.075643*** 0.01366
0.2669*** 0.05950 0.21203*** 0.06192 0.206571*** 0.06009 0.204601*** 0.06103 0.204601*** 0.06073
0.33050** 0.14175 0.3540371** 0.14432 0.3382354** 0.13274 0.3382354** 0.13259
0.42099*** 0.12833 0.4248101*** 0.12988 0.432523*** 0.12397 0.432523*** 0.12362
0.18091 0.13713 0.18223 0.13734 0.18607 0.13842 0.18607 0.13836
15 - Food and Beverage 0.7232*** 0.26030 0.71026*** 0.26614 0.751684*** 0.26694 0.711206** 0.29691 0.711206** 0.29691
16 – Tobacco 0.02418 0.82794 -0.03451 0.81216 0.33882 0.65368 -0.03164 0.96929 -0.03164 0.96929
17 – Textiles 0.00363 0.27018 0.01253 0.27495 0.06914 0.27509 0.00987 0.31834 0.00987 0.31833
18 - Wearing apparel; dres. & fur dyeing 0.07811 0.32358 0.04936 0.32944 0.05978 0.33514 0.05070 0.40304 0.05070 0.40304
19 – Leath.; lug., hand., harness & foot. 0.23219 0.40501 0.20389 0.41157 0.21189 0.40652 0.20549 0.40691 0.20549 0.40691
20 - Wood & wood & cork pds.(e/ furn) 0.44583 0.40143 0.44566 0.40438 0.55907 0.38317 0.44638 0.42141 0.44638 0.42141
21 - Paper and paper products 0.48121 0.37509 0.48060 0.37523 0.628265* 0.37476 0.47741 0.41728 0.47741 0.41727
22 – Publ., print. & repro. of rec. media -0.30206 0.31004 -0.34082 0.31511 -0.21068 0.30978 -0.33960 0.35744 -0.33960 0.35744
23 - Coke, ref. petr. pds.& nuclear fuel -1.02409 0.88759 -1.07877 0.88458 -0.90057 1.05775 -1.08118 0.71890 -1.08118 0.71889
24 - Chemicals 0.06796 0.33194 0.03639 0.33633 0.21604 0.34548 0.03896 0.33284 0.03896 0.33283
25 - Rubber and plastics products 0.20163 0.30570 0.15173 0.31041 0.33979 0.33187 0.15426 0.36554 0.15426 0.36553
26 - Other non-metallic mineral pds. 0.08991 0.27988 0.06515 0.28778 0.31582 0.28168 0.06661 0.35182 0.06661 0.35182
27 - Basic metals -0.56774 0.42532 -0.58615 0.43247 -0.53719 0.42111 -0.58657 0.41974 -0.58657 0.41974
28 – Fabric. metal pds. (e/mach.& eq.) -0.36279 0.31747 -0.36343 0.32394 -0.22647 0.31601 -0.36176 0.35928 -0.36176 0.35928
29 - Machinery and equipment n.e.c. 0.10686 0.28370 0.07442 0.28946 0.16718 0.29133 0.07829 0.32552 0.07829 0.32550
30 - Office, acc. & computing mach. -0.58553 0.71765 -0.57695 0.79432 -0.61578 0.85656 -0.57157 1.14792 -0.57157 1.14791
31 – Electr. mach. & apparatus n.e.c. -0.62531 0.39817 -0.64780 0.40029 -0.53794 0.40330 -0.64756 0.38861 -0.6475601* 0.38861
32 - Radio, TV and comm. eq. -0.65906 0.48209 -0.71598 0.49014 -0.25722 0.56264 -0.71390 0.59498 -0.71390 0.59498
33 – Med., prec. and opt. inst., watches & clocks
0.52313 0.36550 0.51263 0.36654 0.45258 0.36217 0.51026 0.52661 0.51026 0.52661
34 – M. vehic., trailers & semi-trailers 0.46818 0.31657 0.43973 0.31865 0.714635** 0.33567 0.44330 0.37303 0.44330 0.37302
35 - Other transport equipment -0.19152 0.52051 -0.20178 0.52491 -0.21276 0.53650 -0.20635 0.52850 -0.20635 0.52848
11.25807*** 0.20671 11.26248* 0.20468 11.29211*** 0.20738 11.26102*** 0.11446 11.26102*** 0.11445
0.000001* 0.00000 0.000001** 0.00000 0.000001** 0.00000 0.00000 0.00000 0.000001*** 0.00000
1.210392*** 0.18226 1.21030*** 0.18172 1.212478*** 0.17037 1.212478*** 0.17036
-0.65341*** 0.05934 -0.67072*** 0.05943 -0.674503*** 0.05945 -0.67347*** 0.06077 -0.67347*** 0.06073
-0.3222 0.21540 -0.51236** 0.22870 -0.606456*** 0.22809 -0.51826* 0.28126 -0.518263* 0.28122
Individual heterogeneity
0.11600*** 0.01774 0.1156176*** 0.017723 0.1276349*** 0.01829 0.118866*** 0.01707 0.118866*** 0.01686
0.000001*** 0.00000 0.000001*** 0.00000 0.000001*** 0.00000 0.000001*** 0.00000 0.000001*** 0.00000
1.29749 1.30199 1.34709 1.38400 0.04696
1.07299 1.06526 1.06526 1.06901 0.02708
0.59386 0.59901 0.61526 0.626325*** 0.02205
Within 0.8893 0.8915 0.8913
Between 0.9448 0.9446 0.9411
Overall 0.9389 0.9389 0.9359
Obs.: 2400. ***, ** and *: significant at 99%, 95% and 90%, respectively. =total exports, annual average for the sub-period, ARS 000 in constant prices 1998, natural logarithm. RE: random effect, robust standard errors. MLE: maximum-likelihood random-effect. PA: GEE population averaged, convergence not achieved.
time-average value of the corresponding variable. omitted due to collinearity, = 2002-2004. Source: own elaboration based on INDEC (2010).